TABLE OF CONTENTS

The National Digitalker

RS232C Line Driver and Receiver Chips

The Signetics NE566 VCO Chip

Transmission-in-Progress Alarm

The Smith-Kettlewell
Auditory Breakout Box

THE NATIONAL DIGITALKER
By Tom Fowle

Abstract

Use of the "National Digitalker," a chip set which produces high-quality, limited-vocabulary digitized speech
output, is described. Full details on pin connections, control,
and vocabulary are included.

The "National Digitalker" is a set of from two to five
chips, which when properly controlled and addressed, can produce
a very high quality reproduction of "digitized" human speech.

The 40-pin controller chip (MM54104) is powered from a 7 to
11 VDC supply, and controls from one to four ROM's (read-only
memories) containing the digital representation of the vocabulary. The audio output pin of this controller is taken through a
voltage follower, a simple R.C. filter, and an audio amplifier to
drive a speaker.

Using the basic set of two ROM's, MM52164SSR1 and
MM52164SSR2, a vocabulary of 143 words is available, consisting
of the entire alphabet, the numbers "one" to "ten," "twenty" to
"hundred," "thousand," and "million." The rest of the vocabulary
is made up of a set of words having, more or less, industrial
applications -- including the common electrical terms, such as
"volt," "ampere," etc. Using the full set of four ROM's (which
requires additional circuitry), you get 250 words total. Unfortunately for designers of talking computers, this does not easily
make up the ASCII character set, but who wants a computer that
talks mostly in spelled speech, anyhow?

The full numbers-vocabulary ROM, and both sets of two additional ROM's, is given later, but due to space considerations and
the tendency to feel snowed with too much information, the circuitry for connecting all four ROM's is not included. This needs
an extra control line and four more chips, and editorial opinion
is that most applications we might run into wouldn't justify the
extra complexity and expense.

The Digitalker is controlled by eight address and two control lines. A binary representation of the address associated
with the desired word is impressed upon the eight address lines,
and the "Write/Not" line is brought to logic low. This loads the
desired address into the controller. When this line is again
brought to logic high, speech begins, and the "Interrupt" line is
taken low by the controller. When speech is finished, the
"Interrupt" line is brought high again, alerting the system
driving the Digitalker that it is ready for another word address.
This "Interrupt" is only a signal; nothing will stop you from
loading a new address and starting a new word while the previous
word is still being spoken, except that the speech will be interrupted.

If the Digitalker is being driven by a computer, and if you
are into the gory details of programming "exact timing loops,"
you might be able to build new words using parts of words already
in the vocabulary. This might be cheaper than National's price
of $400 per new word (this is if you are a really hot programmer
and you can justify the time). (I haven't become this desperate
yet, and I get paid for my programming time.) The vocabulary
includes such extras as "s," "re," and several silences, making
it possible to mix and match words to form phrases and arrange
the timing of words for clarity. The resultant speech is of very
high quality, good enough so that the naive listener usually has
little trouble understanding it.

Unfortunately, the speed of the speech cannot be changed,
except by varying the frequency of the controller's external
crystal clock; this also changes the pitch of the speech, leading
to the "Donald Duck" effect. However, since the Digitalker
starts out at a male baritone, there is some leeway for this, and
some users have chosen a clock frequency a bit higher than the
prescribed 4mHz with good results. Since the pulse width of the
signal from an external clock, which can be connected to pin 1 of
the controller chip, is specified as no less than 100 nanoseconds, you are limited to a maximum frequency of 5mHz, assuming
you've got a really "squarewave" clock.

Circuit Considerations

On the controller chip, pin 40 is VCC, while pin 20 is
ground. VCC for this chip is between 7 and 11 VDC, and pin 40 is
bypassed to pin 20 by 0.1uF. Maximum current is listed at 45mA.

Pin 3 is called "Chip Select Not," and can be taken high to
"open" the input address and control lines. This is used in
cases where the Digitalker is connected to a computer bus, and
the address lines need to be floated while the bus is doing
something else. In other words, taking pin 3 high makes the
Digitalker turn a deaf ear to all of its inputs.

Pin 4 is "Write Not," and, as mentioned before, is brought
low to load an address into the controller, then brought high
again to start speech. In other words, this is the pin by which
you "trigger" the Digitalker.

Pin 5 is "Not ROM-Power Enable," an output which can be used
to control the power to the ROM's. This is used in cases of
battery supply where current drain is important; the ROM's will
have their power controlled by the controller.

Pin 6 is the "Interrupt Output," (equivalent to the "Busy
Line" of the old TSI Speech Board); this line goes low when an
address is loaded into the chip, then goes high again when speech
is finished. This signal can be used to control the driver
circuitry (or other controlling device), in which case it tells
the driver to "Hold the phone!" while the speech is running.

Pin 7 is called "CMS," and its state controls the action of
the "Write Not" line. With pin 7 low, the operation of pin 4 is
as described. If pin 7 is brought high, raising pin 4 high after

loading an address serves only to reset the interrupt and does
not start speech. This facility is probably intended for use
where the interrupt line really controls the hardware interrupt
of a computer, and where the program taking care of the interrupt
may not have another word to say every time the Digitalker is
finished. I have found no particular use for pin 7, and I simply
ground it for normal operation.

Pins 8 through 15 are the eight input address lines, with
pin 8 being the most significant BIT and pin 15 being least
significant. These address lines are "active high." They should
never be left open. They are TTL-compatible; this means that
logic low is ground and logic high is plus 5VDC. (Actually,
being MOS inputs, you can take them as high as the VCC on the
controller, but a 5V supply is required for the ROM's anyhow --
it's there if you want to use 5V.)

If you want to build up a Digitalker -- not to be controlled
by other logic circuitry, but by switches -- you will need to run
each of the eight address lines to ground through a "pull-down"
resistor of 47K or so; then run each line through a normally-open
switch to plus 5 volts. Closing a switch brings that line to
logic 1. I did this using eight DIP switches, but toggles might
be more fun, though more expensive. Then again, you could be
fancy and use "hex" switches, two being needed, which have a
common swinger and four digit lines. These little items have 16
positions, and give the binary representations of the numbers 0
through 15 on the four lines, with the common (or swinger) being

taken to plus 5VDC in this case.

Pins 16 through 24 are the eight data lines which bring data
from the ROM's to the controller, with pin 16 being called "ROM
Data 1," and pin 24 being "ROM Data 8."

Pins 25 through 38 are the thirteen address lines which
select location in the ROM's to be read by the controller. Pin
25 is "Address 0," pin 37 is "Address 12." Pin 38 is called
"Address 13," and its use is interesting, as you shall see.

These lines which interconnect the controller with the ROM's
are taken to the same pin on each of however many ROM's you use,
from one to four. The proper interconnection of these pins is
given in the "wire-wrap table" that appears later. At first it
was hard to see how the two ROM's could be differentiated until
we discovered that on each set of two ROM's, pin 38 of the controller goes to a "Chip Select" on one, and a "Chip Select Not"
on the other. (As can be seen from the table, pin 38 of the controller goes to pins 20 on all the ROM's.) Thus when pin 38 is
high, one of each set of two chips is selected, and when 38 is
low, the other is active. As said before, using four ROM's is
harder, and won't be covered here for space-and-fear considerations.

As mentioned, there will be a pin table which gives the
interconnection of these pins by numbers only; this is a common
way of describing digital circuitry, where there are only chips
involved, and all you need to know is which pins go to which.

After the pin table, you will find the vocabularies for the
three sets of ROM's. The binary code for each 16th word is
given, so if you can count from 0000 to 1111 in binary, and count
words between the stated addresses, just replace the four trailing 0's with your count; set that 8-BIT address into the eight
input lines, and with an audio system and power in place, bring
"Write Not" low and high again. No, someone didn't sneak up and
sit on your bench, that's Digitalker.

Pin 39 is the audio output. National provides two filtering
circuits, of which I have tried only the simpler, as it works
fine. I have decided not to include the more complex filter
here; it seems unnecessary, and uses a quad op-amp and many
sections of R.C. filtering.

Audio Circuit (simpler, but quite adequate, filter amplifier)

The pin 39 output is buffered by a follower, made up of
just about any old op-amp, say a 741. Thus pin 39 of the Digitalker goes to pin 3 of the op-amp; pins 2 and 6 of the op-amp

are tied together. Pin 4 of the op-amp is grounded, while pin 7
goes to VCC (along with pin 40 of the controller). Pin 6 of the
op-amp, the output of this follower, goes through 9.1K, then
through 0.1uF to ground. The junction of this resistor and cap
goes through 10K to the top of a 50K volume-control pot whose
bottom is at ground.

The wiper of this volume-control pot is taken through 0.1uF
to pin 3 of the LM386 audio amplifier. The 386 has pins 2 and 4
grounded, with pin 6 going to VCC. Pin 6 is bypassed to ground
by 100uF (plus at pin 6). Pin 7 goes to ground through 25uF
(plus at pin 7). Pin 8 goes through 1K to the negative side of a
10uF cap whose plus lead goes to pin 1. Pin 5 goes through
0.22uF to pin 4 (close to chip). Pin 5 also goes to the plus of
a 220uF cap whose negative end goes to one side of the speaker.
The other side of the speaker is grounded.

Pins 1 and 2 are the connections to the crystal oscillator
whose circuit will be given presently. Pin 2 is the drain of an
FET, and pin 1 is the gate. Without the crystal oscillator
described, an external clock signal can be put into pin 1. They
specify a required voltage swing from below 1.2V (logic low) to
above 5.5V (logic high). The minimum pulse width (for both the
high and low halves of the cycle) is 100 nanoseconds.

Crystal Oscillator Circuit

Pin 1 of the controller chip goes
through 20pF to ground, and pin 1 also goes through 1 megohm in
series with 1.5K to pin 2. The junction of the 1meg and the 1.5K
goes through 50pF to ground. The 1meg resistor is shunted by the
4mHz crystal. (The specified crystal is available from Jameco as
part No. CY4A.)

The ROM's are powered from plus 5V, which can be obtained
from somewhere else in the circuit if it is available. However,
it is common practice to put a 5V regulator on the same board as
the Digitalker, powering the regulator, the controller and audio
system, from the same 7 to 11 volt source.

The "Common" terminal of the regulator is grounded. The
"Input" to the regulator, a uA7805, goes to VCC, and it should
probably be bypassed by 100uF or so; its output should of course
have 0.1uF going to ground. This 0.1 might very well be paralleled by 10uF or so just to keep ROM noise off the system.

On all the various ROM's (which are 24-pin chips) pin 24 is
VCC (plus 5V), pin 12 is ground, and (you guessed it,) pin 24 is
bypassed to pin 12 by 0.1uF, close to the chip.

Wire-Wrap Table

As mentioned before, all the ROM's are in parallel; that is,
in sets of two. If you really want to use all four ROM's at
once, I guess you'll have to complain to the author or editor and
see if you can get the necessary additional circuitry out of me.
This would almost certainly mean you are going to run the beast
with a computer, as the hard-wired logic to control 250 words
otherwise would make our old talking meter circuits look small.

Probably the single "numbers" ROM will be enough for most
applications, since it contains the numbers and a "point." This
should do nicely for meters and so forth. The only lack here is
a minus sign, and I suggest you try a few illegal addresses; they
make strange noises, and you can probably find one which makes an
appropriate bleep or buzz for a minus sign. Try them anyhow; no
matter what ROM's you use, they are a real kick, and include one
of the greatest "spring" sounds I've ever heard.

This is a good wire wrap project, and if you have a wrapping
tool which can "chain," or make more than two connections without
breaking the wire, the hookup of this beast with two ROM's will
be a lot easier. Don't worry about what these pins do as you
wire them up, just count pins. Do so carefully! Count in one
direction, then recount pins from the other end of the socket.
Then, check your work against the wire-wrap table with a continuity tester. Retesting can rarely be overdone, especially when
wire wrapping. If you think it's hard the first time, just hope
you don't have to correct many mistakes. Try it -- it's a great
way to kill time till the next good book comes along, and you
will really like the finished project.

In using a pin table, sighted folks mark completed connections on the chart with different-colored ink so they don't

forget what is done. Try moving a paper clip or a small paperweight around the page; it's probably a mess, but otherwise
you'll have to retest everything every time you leave the job for
a cup of coffee -- or just to go scream.

In this table, the first number of each set of two numbers
is the pin on the 40-pin controller, and the second is the pin on
one and/or two ROM's to which it is to be hooked. For example,
you first see 16-9. This means that pin 16 on the controller is
connected to pin 9 on one or both ROM's, depending on how many
you have. In other words, when two ROM's -- of a set of two --
are used, they are wired in parallel.

16-9; 17-10; 18-11; 19-13; 21-14;
22-15; 23-16; 24-17; 25-8; 26-7;
27-6; 28-5; 29-4; 30-3; 31-2;
32-1; 33-23; 34-22; 35-19; 36-18;
37-21; 38-20.

ROM Vocabularies

Numbers ROM

This is the simplest ROM, and perhaps the
most useful, especially if you want just numbers for your talking
digital windmill -- or whatever. It is a single ROM containing:
the numbers from "0" through "9," a "point," as well as three
measured silences to be used for spacing numbers out over time.
For projects containing hard-wired logic, this is definitely the
ROM set of choice, since the others do not say "zero" for an
address of 00000000; instead, they waste this address on advertising.

Address 00000000 (remember, this is in binary) is the word

"zero." Address 00001001 is "nine," and 00001010 is "point."
The next three are the silences (short to longer, respectively),
ending with the address of 00001101. The rest of the possible
addresses are not used. You won't hurt anything by trying them,
and they often make interesting noises. The National's part
number for this "numbers ROM" is MM52116SHRL, and Jameco's catalog number is DT1055.

The first set of multiple ROM's has all the letters, numbers, and many commonly used words. National's part numbers for
this set are MM52164SSR1 and SSR2. Jameco has them as DT1053. A
1kit including these ROM's and the controller is DT1050. This set
costs about $35.

One annoying feature of this set is that "0" is not at an

address of 0; a little advertisement is spoken instead; namely,
"This is Digitalker" is spoken in a female voice, whereas the
rest is in a male baritone. (Chauvinists.)

The vocabulary for this set, beginning with a binary address
1, is given below. To save space, the table is broken up into
groupings; the binary number heading each group is that which
gives you the first word in the group. For example, the first
group (speaking just numbers) starts with an address of 00000001,
which sets this "chip set" up for speaking "1." In the third
group, for example, an "A" is spoken for the address of 00010000;
by counting in binary, you can see that the letter "J" would be
gotten with an address of 00101010.

• 00000001:
  • 1; 2; 3; 4; 5; 6; 7; 8; 9; 10; 11; 12; 13; 14; 15.
• 00010000:
  • 16; 17; 18; 19; 20; 30; 40; 50; 60;
    70; 80; 90; hundred; thousand; million;
    zero.
• 00100000:
  • A; B; C; D; E; F; G; H; I; J; K; L; M;
    N; O; P.
• 00110000:
  • Q; R; S; T; U; V; W; X; Y; Z; again;
    ampere; and; at; cancel; case.
• 01000000:
  • cent; parenthesis; 400Hz tone; 80Hz
    tone.
• 01000100:
  • silences (in milliseconds) of 20; 40;
    80; 160; 320.
• 01001001:
  • centi; check; comma; control; danger;
    degree; dollar; down.
• 01010000:
  • equal; error; feet; fuel; flow;
    gallon; go; gram; great; greater;
    have; high; higher; hour; in; inches.
• 01100000:
  • is; it; kilo; left; less; lesser; limit; low; lower; mark; meter; mile; milli; minus; minute; near.
• 01110000:
  • number; of; off; on; out; over;
    parenthesis; percent; please; plus;
    point; pound; pulse; rate; re; reader.
• 10000000:
  • right; ss; second; set; space; speed;
    star; stop; than; the; time; try; up;
    volt; weight.

The second set of multiple ROM's available -- and probably
the least useful, unless you have some complex industrial process
to monitor -- is the MM52164SSR5 and SSR6. The Jameco number for
this set is DT1057. Its vocabulary follows:

• 00000000:
  • abort; add; adjust; alarm; alert; all;
    ask; assistance; attention; break;
    button; but; call; caution; change;
    circuit.
• 00010000:
  • clear; close; complete; connect;
    continue; copy; correct; date; day;

    decrease; deposit; dial; divide; door;
    east; ed.

• 00100000:
  • ed *1; ed *2; ed *3; emergency; end;
    enter; entry; er; evacuate; exit;
    fail; failure; farad; fast; faster;
    fifth.
• 00110000:
  • fire; first; floor; forward; from;
    gas; get; going; (oh really?) half;
    hello; help; hertz; hold; incorrect;
    increase; intruder.
• 01000000:
  • just; key; level; load; lock; meg;
    mega; micro; more; move; nano; need;
    next; no; normal; north.
• 01010000:
  • not; notice; ohms; onward; open;
    operator; or; pass; per; pico; place;
    press; pressure; quarter; range;
    reach.
• 01100000:
  • receive; record; replace; reverse;
    room; save; secure; select; send;
    service; side; slow; slower; smoke;
    south; station.
• 01110000:
  • switch; system; test; th; thank;
    third; this; total; turn; use; uth;
    waiting; warning; water; west; which.
• 10000000:
  • window; yes; zone.

*1 This is the "ed" used with words ending
in T or D.

*3 This "ed" is used with words ending in
soft sounds such as "k" in the word
"ask."

*2 This "ed" is used with all other words.

As mentioned before, the addresses beyond those listed in
the vocabularies are not used, and produce strange and entertaining noises. These are especially useful for causing laughter
when you have made a wirewrapping error.

As you can see, this system can be controlled with just nine
switches, or from a computer -- or anything in between.

If you build up a Digitalker in which the "Write Not" line
is controlled by a switch, notice that the switch bounce may
cause repeated starting of the word. This is no big deal, and it
will not happen when a proper pulse is delivered.

In the future, we hope to present some sort of talking meter
interface using a Digitalker, but it doesn't look as though it
will ever be very easy to make your digital "whatever" talk. In
a lot of cases, it may be easier and cheaper to go from scratch
rather than trying to modify existing digital-output gear.

You might find it interesting to note that many of the so-
called speech synthesizer systems being sold for the simpler home
computers use this chip set. They are not really speech synthesizers, as they do not create speech, just play it back. A
future article will discuss this issue: what is a synthesizer,
and which things are just solid-state, random-access
recorder/player systems. They both have their uses, and we must
learn which fits where.

Review of Digitalker Hookup

Just to have it all in one spot, here is an example circuit,
all-inclusive:

No matter how many ROM's you have, their pin 12 is grounded
and 24 goes to plus 5V. On the controller, pin 20 is grounded,
while pin 40 goes to a VCC supply of 7 to 11 volts. In both
cases, located close to the chips, the supply lines are bypassed
to ground by 0.1uF.

The 5V is gotten from a uA7805 regulator. Its "Common"
terminal is grounded. Its "Input" goes to VCC (and to pin 40 of
the controller). This "Input" is bypassed by 100uF (negative at
ground). The 5V "Output" is bypassed by the parallel combination
of 0.1uF and 10uF (negative of the electrolytic at ground).

The controller and its ROM's are interconnected as follows:

16-9; 17-10; 18-11; 19-13;

21-14; 22-15; 23-16; 24-17;
25-8; 26-7; 27-6; 28-5; 29-4;
30-3; 31-2; 32-1; 33-23; 34-22;
35-19; 36-18; 37-21; 38-20.

Pin 39 goes into an audio system as follows:

Audio Circuit (simpler, but quite adequate, filter amplifier)

The pin 39 output is buffered by a follower, made up of
just about any old op-amp, say a 741. Thus pin 39 of the Digitalker goes to pin 3 of the op-amp; pins 2 and 6 of the op-amp
are tied together. Pin 4 of the op-amp is grounded, while pin 7
goes to VCC (along with pin 40 of the controller). Pin 6 of the
op-amp, the output of this follower, goes through 9.1K, then
through 0.1uF to ground. The junction of this resistor and cap
goes through 10K to the top of a 50K volume-control pot whose
bottom is at ground.

The wiper of this volume-control pot is taken through 0.1uF
to pin 3 of the LM386 audio amplifier. The 386 has pins 2 and 4
grounded, with pin 6 going to VCC. Pin 6 is bypassed to ground
by 100uF (plus at pin 6). Pin 7 goes to ground through 25uF
(plus at pin 7). Pin 8 goes through 1K to the negative side of a
10uF cap whose plus lead goes to pin 1. Pin 5 goes through
0.22uF to pin 4 (close to chip). Pin 5 also goes to the plus of
a 220uF cap whose negative end goes to one side of the speaker.
The other side of the speaker is grounded.

An external 4mHz clock can be fed into pin 1 (having a
peak-to-peak voltage of 6V or so). On the other hand, a crystal
oscillator can be built around pins 1 and 2 as follows:

Crystal Oscillator Circuit

Pin 1 of the controller chip
goes through 20pF to ground, and pin 1 also goes through 1 megohm
in series with 1.5K to pin 2. The junction of the 1meg and the
1.5K goes through 50pF to ground. The 1meg resistor is shunted
by the 4mHz crystal. (The specified crystal is available from
Jameco as part No. CY4A.)

Pin 3, the operation of which deafens the inputs, is grounded for normal use. Pin 7, which changes the function of pin 4
(see text) is grounded for normal operation.

Pins 8 through 15 are the "Address Inputs"; a binary number
is to be impressed on them. Pin 8 is the most significant BIT,
and pin 15 is the least significant.

Now, bringing pin 4 low loads the address; bringing it high
again makes it talk. During this "busy" process, pin 6 (the
"Output Interrupt" or "busy line") can be observed with a voltmeter or a logic probe.

Suppose we want to test the chip set, or make a good party
gimmick. Run pins 8 through 15 all through 47K resistors to
ground; put toggle switches between them and 5V (or VCC, at your
convenience). Next, run pin 4 through 47K to plus 5V (or VCC),
and connect a normally-open pushbutton "start" switch between pin
4 and ground.

"This is Digitalker."

_________________

Supplier Address:
Jameco Electronics, l355 Shoreway Rd.,
Belmont, CA 94002 (4l5) 592-8097

RS232C LINE DRIVER AND RECEIVER CHIPS

Abstract

Whereas the logic systems in computers (their "1's" and
"0's") are operated from a single supply, usually 5V, the levels encountered
in the EIA standard RS232 serial port systems are bipolar, going from a minus
voltage to a plus voltage. In addition, to make these ports immune to noise,
a 6-volt "dead band" is centered at 0V (from 3V below to 3V above ground).

Finally, transmitted signals are current-limited (at 10mA in the case of these
chips); if a line gets shorted out, the cable won't burn up and the computer
equipment won't be damaged. While all of these are excellent ideas, it does
mean that "line-driver" and "receiver" chips are needed to interface between
single-ended logic levels and RS232C serial systems.

The Quad Driver Chip
TI SN75188 or Motorola MC1488

A 14-pin package, this chip contains four drivers. Three of these are
NAND gates, while the fourth is an inverter. Their inputs are compatible with
TTL logic (with some additional forgiveness as to input levels). Powered from
a split supply (from plus/minus 9V to plus/minus 15V), the outputs jump between these extremes -- minus for an input logic high, and plus for an input
logic low.

The EIA standard for RS232C levels says that the voltage can go as high
as plus/minus 25V. For this reason, the literature suggests that this device
be powered through diodes in the supply lines. These diodes would prevent the
outputs of the chip from sinking current if, for some reason, an output were
pulled beyond the supply voltages externally.

Specifications for the Driver Chip

• Supply Voltages--plus and minus 15V
  supplies, maximum
• Supply Current for plus/minus 15V Supplies
  --34mA maximum from each supply,
  measured with all inputs high
• Input Voltage Range--from minus 15V to plus
  7V
• Input Logic High--1.9V
• Input Logic Low--0.8V
• Output levels for plus/minus 9V Supply--
  Typically plus/minus 7V, with minima
  of 6V
• Output Levels for plus/minus 13.2V Supply--
  Typically plus/minus 10.5V, with minima
  of 9V.
• High-level input Current--10uA
• Low-Level Input Current--minus 1.6mA
  maximum.

Pin Connections

• Pin 1--VCC minus
• Pin 7--Ground
• Pin 14--VCC plus
• Pin 2--Inverter input
• Pin 3--Inverter output
• Pins 4 and 5--Inputs to first NAND
• Pin 6--Output of first NAND
• Pins 9 and 10--Inputs to second NAND
• Pin 8--Output of second NAND
• Pins 12 and 13--Inputs to third NAND
• Pin 11--Output of third NAND

Applications

Because the outputs are current limited, the literature shows some novel
uses for these devices -- shifting levels from TTL to other logic types. For
example, one setup is for driving RTL logic whose preferred levels are from
minus 0.7 to plus 3.7 volts. They do this with diode clamps on the outputs;
one diode (anode grounded) keeps the output from going below minus 0.7V, while
another diode (cathode to a 3V supply) keeps the output below 3.7V. Use this
principle freely, recognizing that the input can go from minus 15V to plus 7V,
and that the outputs can be clamped to whatever you wish.

We used this chip to build a test jig by which various prototype breakout boxes and transmission detectors were tested before we hooked them up to
computers. Since the circuit of this test jig is rather typical, it is detailed here.

RS232C Test Generator Circuit

A commercial dual supply (plus/minus
15V) was used, with the common terminal of this unit grounded. In addition,
the plus 15V line goes to the input of a 5V regulator, this input being bypassed by 100uF (negative at ground). The common terminal of the regulator is
grounded. Its 5V output goes to a 5V line which is bypassed by another 100uF
(negative at ground). (This 5V line is for an input clock and other input
voltages; it is not necessary for powering the RS232 driver chip.)

Pin 7 of the driver chip is grounded. Pin 1 goes to the anode of a
diode (1N4001), with the cathode of this diode going to minus 15V. Pin 14
goes to the cathode of another diode, with the anode of this one going to plus
15V.

Pins 12 and 13 of the driver chip are grounded; this makes pin 11 always
high. Pins 9 and 10 of the driver chip go to plus 5V; this makes pin 8 always
low.

Pin 2 goes to the arm of an SPDT switch; a position marked "low" goes to
plus 5V, while a position marked "high" is grounded. Pin 3, the inverter's
output, does as the switch says.

The data-faker clock uses a 555 timer. Pin 1 is grounded, while pin 8
goes to plus 5V. Pins 2 and 6, which are tied together, go through 0.01uF to
ground. Pins 2 and 6 also go through 6.8K, then through a 250K rheostat to
the output, pin 3. This pin 3 goes to pins 4 and 5 of the driver chip.

Pin 4 of the 555 goes to the arm of an SPDT "data on-off" switch. The
"off" position is grounded, while the "on" position goes to the 5V line. When
this switch is "off," the pin 6 output of the driver chip is low (at minus V);
when this switch is "on," pin 6 generates a bipolar squarewave.

Quad Receiver Chip
TI SN75189 or Motorola MC1489

This chip contains four inverters. The customary bipolar input signals
are converted to 5V TTL single-ended signals. Of course, being inverters, a
negative input signal is converted to a 5V output signal, while a positive
input signal sends the output of its inverter to ground.

Although this chip's inputs "understand" and tolerate the bipolar signals to be expected on RS232 lines, it does not exactly replicate the constraints of official standards. For example, there is no undefined region (the specified 6V dead band) about ground; rather, each inverter is a comparator with hysteresis that will be "overcome" by signals of the proper level. A
resistor off a "control pin" allows you to choose the voltage about which each
comparator operates. In addition, bypassing the control pin suppresses noise
pulses; the height and duration of pulses to be ignored varies with the choice
of this capacitor.

The above deviation from the EIA standard makes the chip useful for
other than RS232 applications. One can envision uses for a 5V chip whose
inputs can be taken above and below the supply (up to plus and minus 30V).
One should remain cognizant of the fact that the input impedance can be as low
as 3K, however.

Besides the supply pins (for plus V and ground), each receiver in the
chip has three pins: "Input," "Output," and "Control." Noise suppression can
be accomplished by bypassing the "Control" pin to ground. The mean voltage
about which the comparator switches can be shifted by running the "Control"
pin through a resistor to an external voltage source (shown to be plus or
minus 5V in exemplary circuits).

Specifications

• Supply Voltage--10V max.
• Supply Current--26mA max.
• Package Dissipation--1 watt max.
• Input Voltage Maxima--plus/minus 30V
• Input Impedance--3K min., 7K max.
  (Note: The following are given for a
  supply voltage of 5V.)
• High-Level Output Voltage--4V typ.
• Low-Level Output Voltage--0.2V typ.
• Short-Circuit Output Current--3mA typ.
• Positive-Going Threshold Voltage--1V min.,
  1.5V max.
• Negative-Going Threshold Voltage--0.75V min.,
  1.25V max.

Noise-Rejection Specifications

The literature shows a family of curves, one curve for each capacitor
value. Each curve, plotting pulse amplitude vs. pulse width, shows the range
of noise pulses which will be rejected with a given bypass capacitor on the
control pin. (The capacitor goes from the control pin to ground.) Some
sample values are given below:

With 10pF, the device will ignore a 6V, 10-nanosecond pulse, as well as:
3V for 30ns, 2V for 70ns, finally asymptotically approaching 1.25V for 1000ns
or longer. With 100pF, the following pulses will be rejected: 6V for 80ns,
3V for 160ns, 2V for 300ns, and so forth. With 500pf: 6V for 250ns, 4V for
300ns, 3V for 400ns, 2V for 850ns approaching 1.25V at 10,000ns and greater.

It makes sense that all the curves converge as they approach 1.25V. In
other words, if you keep the input voltage above the positive-going threshold
(1.25V), the comparator will trip eventually. The bypass capacitor just makes
the system slow to respond.

Shifting the Input Threshold Levels

The total hysteresis of each comparator is nominally about 0.4V. This
amount of hysteresis never changes; the threshold for a positive-going voltage
will be 0.2V above the "mean," while the threshold for a negative-going voltage will be 0.2V below the mean. What you can do, by manipulating the control
pin, is change the position of the mean about which the comparator trips.
Some examples are given below:

1. With nothing on the control pin, the
   mean is just above 1V (perhaps
   1.125V).
2. With the control pin going through 5K
   to a plus 5V source, the mean drops
   down to minus 1.75V.
3. With the control pin going through
   13K to plus 5V, the mean is at about
   0V.
4. With the control pin going through
   11K to minus 5V, the mean is about
   2.7 or 2.8 volts.

Pin Connections
SN75189 and MC1489

• Pin 7--Ground
• Pin 14--VCC
• Pin 1--Input 1
• Pin 2--Control 1
• Pin 3--Output 1
• Pin 4--Input 2
• Pin 5--Control 2
• Pin 6--Output 2
• Pin 10--Input 3
• Pin 9--Control 3
• Pin 8--Output 3
• Pin 13--Input 4
• Pin 12--Control 4
• Pin 11--Output 4

THE SIGNETICS NE566 VCO CHIP

Abstract

This 8-pin chip produces both square- and triangle-waves
over a 10-to-1 frequency range, the frequency being controlled by a high-
impedance input pin. Its advantages are that it has low output impedances,
and that it costs less than $1.50 from Jameco. Its disadvantages are that the
output waveforms ride atop oddball bias voltages, and that the control signal
is referenced to VCC, not ground.

This chip has three of the features contained in the ICL8038 (see SKTF,
Fall 1983); it has square and triangle outputs, and a VCO control pin. Why
not use the 8038 and skip this one? Having only 8 pins, it has a size advantage. Also, if you remember the 8038, its output was easily loaded: this
Signetics chip has emitter-followers for outputs, their stated impedance being
50 ohms. Finally, it's about one-third the cost of the 8038, being only about
$1.50.

Another oddity of this Signetics chip is that the output waveforms have
strange biases on them: The squarewave goes from about 1/2 VCC up to VCC,
while the triangle goes from slightly less than 1/3 VCC to 1/2 VCC. The
outputs of the 8038 are centered about 1/2 VCC, which is a little easier to
remember, anyway.

The oscillator circuitry is similar to the 8038; a "programmable current
source" (which is set by choosing a sampling resistor) orchestrates the charge
and discharge rates on a timing capacitor. This current source is also voltage-controlled from the VCO input pin, thus giving you control over the frequency.

Specifications

• Supply Voltage--from 10 to 24V
• Supply Current--7mA typ.
• Maximum Operating Frequency--1 MHz
• Drift in the Center Frequency--200PPM
  per dg. C.
• Control Pin Impedance--1 megohm typ.
• FM Distortion--0.4% with plus/minus 10%
  deviation
• VCO Frequency Range--10 to 1 with good
  linearity
• Triangle Output Impedance--50 ohms
• Triangle Amplitude--typically 2.4V peak-to-
  peak with 12V supply, 1.9V peak-to-peak
  min.
• Triangle Linearity--5% typ.
• Squarewave Output Impedance--50 ohms typ.
• Squarewave Amplitude--typically 5.4V with a
  12V supply, 5V min.
• Duty Cycle--50% typ., 40% min., 60% max.
• Rise Time--20 nanoseconds
• Fall Time--50 nanoseconds

Pin Connections

• Pin 1--Ground
• Pin 8--VCC
• Pin 3--Squarewave Output
• Pin 4--Triangle Output
• Pin 5--VCO Control Pin
• Pin 6--For Frequency-Determining Resistor
• Pin 7--For Timing Capacitor
• Pin 2--No Connection

Note: The package styles you get are determined by suffixes. The usual
plastic package is an NE566V; the NE566T is in a metal can.

Frequency Determination

The formula below gives the frequency as a function of: the voltage on
the VCO control pin, and the external resistor and capacitor which must be
chosen. The operating range of the VCO pin is between 3/4 VCC and VCC. The
frequency goes up as this pin is brought down from VCC, with the maximum being
about twice the center frequency with 3/4 VCC on this pin. (The "center
frequency" is that gotten with the VCO pin at 7/8VCC.) The next constraint is
the resistor R; it is recommended that this be between 2K and 20K.

f = 2*(VCC-V)/(R*C*VCC)

VCC is the supply voltage, V is the voltage on the control pin, R is an
external resistor from pin 6 to VCC, and C is a timing capacitor from pin 7 to
ground.

I usually know what frequency I want. Furthermore, there are restrictions on the resistor's value -- from 2K to 20K. Therefore, a handy expression would be to solve the above formula for the capacitor C as follows:

C = 2*(VCC-V)/(f*R*VCC)

Example

Suppose we want a 1Khz oscillator. We can start by setting
pin 5 to the center of its range, 7/8 VCC. Arbitrarily, let's choose a resistor of 11K. With a supply voltage of 12V, we plug in the numbers as follows:

C = 2*(12-10.5)/(1*10³*11*10³*12)
or C = 2.27*10^-8

Example Circuits

Connection of this chip is very simple, but there are two arrangements
needed in addition to the frequency-determining components. First of all, if

the device is to free-run, a voltage must be applied at pin 5; this can be
done with a voltage divider across the power supply. Second, Signetics recommends that 0.001uF be placed between pins 5 and 6; this is said to be necessary to suppress oscillations in the voltage-controlled current source.

Circuit for a 1kHz Test Oscillator

Pin 1 of the NE566 is grounded,
while pin 8 goes to the 12V line. Between pins 1 and 8 is 0.1uF, located
close to the chip. The 12V line also goes through 12K, then through 82K to
ground. The junction of these resistors goes to pin 5 of the 566. Between
pins 5 and 6 is 0.001uF.

As per our calculations in the previous section: Pin 6 goes through 11K
to the 12V line. Pin 7 goes through 0.022uF to ground.

Besides the above circuit, the literature shows some additional embellishments. For example, the above oscillator can be FM'ed (frequency modulated) by connecting a modulating signal through perhaps 0.47uF to
pin 5.

This chip can be used to drive TTL circuitry only if you run it off of a
split supply -- plus and minus 5V being necessary. This is true, since the
squarewave output is riding on a DC bias. I think a 555 timer or the VCO
section out of the RCA CD4046 PLL chip (See SKTF, Summer 1982) would be a
better choice for logic circuitry.

[Editor's Note: Signetics never really says what the lower limit on the
supply voltage is; their curves all stop at 10V, is all. In my typical
devil-may-care engineering style, I have run these off of 9 volts, and they
seem to work. I cannot guarantee, however, that the performance will hold up
very well as the battery dies.]

A TRANSMISSION-IN-PROGRESS ALARM
FOR MODEM USERS

(This project was initiated on request by Dr. T.V. Cranmer, Chairman of

the Committee on Research and Development, National Federation of the Blind.)

Abstract

When using a talking computer in conjunction with a MODEM,
the user must "ask" his computer whether or not up-loading or down-loading of
information has been completed; this inquiry either interrupts the transmission (which wastes valuable time), or the inquiry might be long overdue (again
wasting time on the phone line). The Transmission-in-Progress Alarm (TIP
Alarm) uses a Mallory Sonalert beeper to monitor activity on pins 2 and 3 of
the standard serial port. A completely passive device, this instrument consists of a modified RS232C connecting cable to which a small box, containing
the sounder, has been attached.

Introduction

Use of "data bases" over a phone line is expensive; this can be more so
(and less efficient) for the blind computer user who has to ask his machine to
read a bit -- read a bit more -- then decide whether or not a transmission is
finished. Expensive MODEMS have a loudspeaker through which the string of
data, and its carrier, can be heard. However, ASCII data FSK is not very
pleasant listening. Thus, there is a need for a "little birdie" in the line
to tell you when pulses appear on the transmit and receive lines (pins 2 and
3).

The TIP Alarm is appallingly simple to build, and it is small enough to
fit into a 1- by 2- by 3-inch plastic box (or any other similar volume). This
box, containing the Sonalert beeper and a volume control, is then taped or
cemented to an appropriate ribbon cable with DB25 connectors on the ends.
(The hardest part about building this instrument is lifting out the wires
assigned to pins 2, 3 and 7 -- then splicing on to them.)

A 1-inch hole in the box accommodates the Mallory Sonalert; the No.
SNP428 (4- to 28-volt miniature unit) was used, although the more common SC628
will probably work as well. You may also wish to provide three "tie points"
on a terminal strip for the two diodes and the resistor; we just soldered them
together and buried the mess in heat-shrinkable tubing. Other than that,
mounting the volume control is all you have left to do.

Circuit

Pin 7 of the RS232C bus is ground by decree; the negative
terminal of the Sonalert is grounded, as is the bottom of the 100K volume
control. The positive terminal of the Sonalert goes to the arm of the volume
control. The top of the control goes through 10K to the cathodes of two
diodes (1N914, 1N4148, etc). One anode goes to pin 2, and the other to pin 3
of the cable.

Steady on Your Pins

The DB25 connector has two rows of pins, 13 on
the top row, and 12 on the bottom. (Always view these connectors with the
long row on top). These pins are staggered such that pin 14 is just below and
between pins 1 and 2. Viewed from the pins of the male connector, the pins
are counted from left to right -- 1 through 13 for the top row, and 14 through
25 on the bottom row (pin 25 is in the lower right-hand corner). As for the

female, these pins are counted from right to left -- 1 through 13 on the top
row, and 14 through 25 on the bottom row (pin 25 is at the bottom left).

If you are looking at a male plug on a ribbon cable, the wires of the
ribbon -- from left to right -- are: 1, 14, 2, 15, 3 . . . Thus, pins 2 and
3 are wires 3 and 5; pin 7 is wire 13. This can be done for the female connector; just count the wires from right to left.

Parts List

• 1--Sonalert, Mallory No. SNP428, Available
  from Mallory Components Group, 3029 E.
  Washington St., Indianapolis, IN 4620l;
  (3l7) 636-5353
• 1--100K volume control, panel-mount
• 1--10K 1/4-watt resistor
• 2--small-signal 50V PIV silicon diodes,
  1N914, 1N4148, etc.
• 1--Appropriate ribbon cable, available from
  Jameco (and everywhere else)
• 1--Small plastic box, perhaps 1- by 2- by 3-
  inches

THE SMITH-KETTLEWELL AUDITORY BREAKOUT BOX

Introduction

In its simplest form, a "breakout box" is an interrupted
cable, having provisions for re-establishing and/or re-arranging the cable's
connections. When two pieces of computer equipment are not "talking to each
other" (meaning that there is a mismatch in the so-called "standard" interface), a breakout box allows the technician to quickly try modifications in
the connecting cable in order to solve the problem.

(Note: At the end of this article, a very brief sketch of the RS232C
port is given -- pin assignments and the like.)

Fancy breakout boxes have indicator lamps (visual logic probes) built
into them, so that the technician can "watch" the signals on the various
lines; "Is there a 'handshaking' signal?", "Is it doing any good?", or "Perhaps I need to reverse the data transmit and receive lines." These lamps
indicate a logic high (red), a logic low (green), or pulses of data (giving
off a blended intermittent yellow).

Our first approach was to build an audible logic probe into a commercial
breakout box (an Inmac No. 369-3). While this scheme worked, it did not
completely satisfy us; there was only enough room for one audible indicator
whose loudspeaker was too small to hear, the jumper pins were not in the same
configuration as the pins on the connector (forcing the user to count them
differently), and all such boxes are justifiably expensive in their visual
form ($250). Why not build from scratch? Our eventual design uses wire-wrap
pins, and appropriate push-on test connectors, for the jumpering system.
Aside from this arrangement (which is completely passive), two battery-operated audible logic probes are included, each of which indicates the following
four conditions:

1. A "logic high" (plus 3 volts or
   higher) is indicated by a high-
   pitched beep.
2. A "logic low" (minus 3 volts or
   lower) is indicated by a beep of
   medium pitch.
3. "Ground" (0 volts) is indicated by a
   low-pitched buzz.
4. An "open circuit" (floating wire) is
   indicated by a low buzz of wavering
   pitch (it has a 4-cycle "wow").

Physical Considerations

The instrument is built into a file-card box
for 3- by 5-inch cards. The logic probes are built on perforated board which
lies on the bottom. Above this circuit board, and mounted on the end aprons
of the box, are two small speakers. A stereo earphone jack is also provided
so as to promote good stereo separation. The breakout box jumper panel is
another piece of perforated board mounted by means of right-angle brackets
near the top of the box. The 9-volt battery is mounted in the lid.

The top panel has two double rows of wire-wrap pins (arranged in staggered configuration, just like the DB25 connectors). A 3-position toggle
between them operates on pins 2 and 3; they are either connected straight
through, reverse-connected (2 to 3 and 3 to 2), or left open for connection by
jumper wires instead. Besides the on-off switch for the logic probes, a 3-
position toggle selects either or both of the audible indicators.

Staggered rows of pins are made using "headers" of the wire-wrap type;
their short pins are fitted into the board, and these headers are glued in
place, leaving their square wire-wrap posts standing up to be used for jumpering. Under the board, their short pins are then soldered to the wires of an
RS232 ribbon cable which has been cut in half. Also under the board, certain
obvious connections between the interrupted cables are made; pins 1 and pins 7
(both ground wires) are jumpered underneath, and the switch operating on pins
2 and 3 is wired up underneath.

[Note: A "header" is merely a plug-like strip of pins; the ones specified here have two unstaggered rows of pins with standard 1/10-inch spacing.
They are often used as connector plugs on the edges of circuit boards. Those
described here have short pins (which could fit into female cable-mount connectors, or which can be soldered to), and they have long square wire-wrap
pins on the other side of their insulator strips.]

These "headers" come in a 50-pin size -- two rows of 25 pins. To make
staggered rows that are laid out like the DB25 units, we simply cut off every

other wire-wrap post; this leaves 13 on the top row (pins 1 through 13), and
12 on the bottom row (14 through 25).

Circuit Operation

The signals on the RS232C interface are bipolar --
they are either below ground by 3 volts or more, or above ground by 3 volts or
more (their extremes being as much as 25 volts, by definition). Ordinary
logic probe circuits won't work at negative input voltages, and it is necessary to design one with these capabilities.

With a zener diode having a pair of resistors in series across it, a
phony ground was created at mid-supply. Either side of this "ground," the
anode of the zener gives us minus 2.8V, and the cathode gives us plus 2.8V --
the voltage levels beyond which we want to call an input a legitimate signal.
Comparators in the LM324 sense these "end point" voltages; the comparators
then drive charging resistors of two audio oscillators in the NE556 timer
chip. Right away, we have taken care of the high and medium tones associated
with the logic levels on the RS232C port.

Now it gets interesting; we felt that it would be nice to know if wires
were truly uncommitted (floating in the breeze), or if they were within the
boundaries of plus/minus 2.8V. In order to make this determination, one could
use a little force -- pulling on the wires with electrical currents as if to
say, "Anybody home?" However, this is impolite, and who knows what you might
affect in doing so. Therefore, we chose to use a buffer (which would subject
the lines to very low currents), whose output replicates its input condition.

A complementary pair of transistors is just such a buffer. This circuit
arrangement is made up of two emitter followers, two complements being used so
that the output can source current in either direction. (A single emitter
follower will maintain its directed output voltage in one direction, but it
acquiesces freely when its emitter is brought toward its collector.) If the
bases of the pair are left floating, the transistors are biased to cutoff, and
you can freely take their emitters anywhere. Connecting the bases to a wire
which is uncommitted preserves the condition of cutoff; any commitment of the
input will cause the emitters to find this voltage and maintain it to the
death. (There is a bit of "sideways play" at the emitters; they can be pulled
0.6V in either direction of the input voltage -- just enough to turn either
one of them on.)

In order to test the condition of the buffer (to see if its input has
caught a fish or not), a modulating signal is imposed on the emitters. This
signal, a 4Hz triangle wave, tries to move the emitters; if it does so, the
bases are uncommitted and the transistors are at cutoff. If the emitters hold
their "ground," the bases are tied to something (most probably, ground). A
condition could exist where the inputs are just above minus 2.8V, whereupon
the triangle could extend partially into the emitters' region of "sideways
play," and some modulation would still occur. However, this is an unlikely
condition, and my advice is not to worry. You could minimize this effect by
increasing the supply voltage to 12 or 15 volts, whereupon the "dead band" of
the emitters would become less significant, compared to the increased degree
of modulation.

The 4Hz triangle wave, which I call "the wobbler," is generated by a
Signetics NE566 function-generator chip. As mentioned in the article describing this chip (contained in this issue), the output waves have odd-ball biases
on them. In order to put the signal where I wanted it (below the dead band of
the buffers when their inputs are at ground), a diode in the negative supply
line was necessary. The outputs of the buffers are then used to influence
their respective audio oscillators via pull-down resistors off the discharge
terminals of the 556 timer chip. If the emitters wobble, the audio oscillators waver in pitch; if the buffers hold steady, a pure low-frequency tone is
heard from them.

Massive decoupling of the various supplies in the circuit was required;
the two halves kept influencing each other until I pulled every trick I could

think of to stop this. I even put bypass capacitors on the 556's "Control"
pins, something I have heard of but never done elsewhere. Even so, cross-
channel effects are not completely gone in my prototypes, although they are
minimal. Although board layout shouldn't make a difference in this regard, I
include a sensible layout immediately following the circuit description.

Breakout Box Circuitry

Power Supply Lines

As mentioned in the text, various decoupling
points and voltage levels are created in this circuit. For this reason, the
following standards must be established:

1. "Minus B"--The negative battery lead.
2. "Circuit Common"--A point which is
   0.7V above Minus B.
3. "VCC1"--The plus battery lead.
4. "VCC2"--Supply for the wobbler.
5. "VCC3"--Supply for the 556 timer

   chip.

6. "V1(standard)"--Cathode of the zener.
7. "V2(standard)"--Anode of the zener.
8. "Ground"--Approximately the midpoint
   of project supply; pin 7 of the
   RS232 port.

The negative battery lead goes to "Minus B," a negative supply point for
the 566 wobbler. This Minus B point also goes to the cathode of a rectifier
diode (1N4001); the anode of this diode goes to "Circuit Common."

The plus battery terminal goes through an on-off switch to "VCC1." This
VCC1 line is bypassed to Circuit Common by 100uF (positive toward VCC1).

The VCC1 line goes through 100 ohms to the "VCC2" line, the latter being
bypassed to the Minus B line by 100uF (positive at VCC2). (This supplies the
566 wobbler.)

The VCC1 line also goes through 10 ohms to the "VCC3" line, with VCC3
being bypassed to Circuit Common by 100uF (positive at VCC3).

VCC1 goes through 220 ohms to the cathode of a 5.6V zener diode (1N752);
The anode of this zener goes through 220 ohms to Circuit Common. The cathode
is bypassed to Circuit Common by 100uF (negative at Circuit Common).

Two 1K resistors are connected in series between cathode and anode of
the zener diode. The junction of these 1K resistors is taken as "Ground," or
"Signal Ground," and goes to pin 7 of the RS232 port.

Wobbler Circuit

Pin 1 of a NE566 function-generator chip goes to
Minus B. Pin 8 of this chip goes to VCC2. Between pins 1 and 8 is 0.1uF
(located close to the chip).

Pin 7 of the 566 goes through 5uF to Minus B (positive toward pin 7).
Pin 6 goes through 12K to VCC2. Located close to the chip is 0.001uF between
pins 5 and 6. Pin 5, the VCO control pin, goes through 12K to VCC2, and
through 82K to Minus B. Pin 4 of the 566 is the wobbler's output (a triangle
wave).

Complementary Pair Input Circuit

The two "channels" are identical.
In each case: The collector of a 2N2907 PNP transistor goes to Circuit Common. The collector of a 2N2222 NPN unit goes to VCC1. Between the emitters
is a 47-ohm resistor. The bases are tied together.

In the case of each input channel, the emitter of the 2907 goes through
560K to Circuit Common; the 2907 emitter also goes through 330K to pin 4 of
the NE566, the output of the wobbler.

Each set of tied-together bases goes to the cathode of a 1N914 diode,
the anode of which goes to Circuit Common. The bases also go to the anode of
another 1N914, with the cathode of this going to VCC1. (In other words, each
pair of bases is clamped to the "rails" by two diodes.) Each pair of bases
then goes through 47K to its channel's input probe.

"High-Low Comparators" Circuits

An LM324 quad op-amp is used to get
two comparators per channel. Pin 11 of this package goes to minus B.
Pin 4 goes to VCC1. Pins 1 and 7 are the "high" and "low" outputs of channel
2, respectively; in the same manner, pins 14 and 8 are the "high" and "low"
outputs of channel 1.

Pins 5 and 10 of the 324 are tied together and go to the anode of the
zener, to the V2(standard). Pins 2 and 13 are tied together and go to the
cathode of the zener, the V1(standard). (These pins represent the non-inverting inputs of the low comparators and the inverting inputs of the high comparators.)

The inverting input of a low comparator and the non-inverting of its
associated high comparator must now be connected so as to sense the input
levels: Pins 9 and 12 of the 324 are tied together and go to the bases of
channel 1. Pins 3 and 6 are tied together and go to the bases of channel 2.

Audio Oscillators

Pin 7 of an NE556 timer goes to Circuit Common.
Pin 14 goes to VCC3. Between 7 and 14, located close to the chip, is 0.1uF.
Pins 4 and 10 (Enable pins) both go to VCC3. Pins 3 and 11 are each bypassed
to Circuit Common by 0.22uF (disc or mylar). (Pins 3 and 11 are "Voltage-

Control pins," and are bypassed just to keep them noise-free.)

Pins 2 and 6 are tied together and go through 0.022uF to Circuit Common.
Between pins 2 and 1 (from Trigger to Discharge) is 22K. Pin 1 also goes
through 220K to VCC3. This pin 1 goes through 560K to the emitter of the
2N2222 transistor associated with Channel 2.

Pins 8 and 12 are tied together and go through 0.022uF to Circuit Common. Between pins 12 and 13, from Trigger to Discharge, is 22K. Pin 13 goes
through 220K to VCC3. Pin 13 goes through 560K to the emitter of the 2N2222
transistor associated with Channel 1.

Pin 1 of the 556 (Discharge of the Channel 2 side) goes to the cathodes
of two diodes (both 1N914's). The anode of one goes through 47K to pin 7 of
the 324 (the associated low comparator), while the anode of the other goes
through 2.2K to pin 1 of the 324 (the high comparator output).

Pin 13 of the 556 (the Discharge of the Channel 1 side) goes to the
cathodes of two diodes (1N914's). The anode of one goes through 47K to pin 8
of the 324 (the associated low comparator output), while the anode of the
other goes through 2.2K to pin 14 of the 324 (the high comparator output).

Pin 9 of the 556, the audio output of Channel 1, goes through 100 ohms,
then through its speaker to VCC3. Pin 5, the audio output of Channel 2, goes
through 100 ohms, then through its speaker to VCC3. The sleeve of an open-
circuit stereo earphone jack goes to VCC3. The tip contact (Channel 1) goes
through 2.2K to pin 9; the ring contact (Channel 2) goes through 2.2K to pin
5. (Plugging in the earphones does not silence the speakers in this arrangement. Why bother; if the noise don't drive you crazy, the computer will.)

Disable Circuit

A single-pole, double-throw switch with a center-off
position is used. The arm of this switch goes to Circuit Common. Position 2,
the Channel 2 Standby position, goes to pins 8 and 12 of the 556. Position 1,
the Channel 1 Standby position, goes to pins 6 and 2 of the 556.

Input Connections

"Signal Ground," the junction of the 1K resistors
across the zener, goes to pin 7 of the DB25 RS232 Connector. As mentioned
before, the Channel 1 test probe goes through 47K to the bases of the complementary pair which also go to pins 9 and 12 of the 324. The Channel 2 test
probe goes through 47K to the transistor bases which go to pins 3 and 6 of the
324. These input probes are made from an alligator clip lead which has been
cut in two. (Actually, we used clips of the spring-loaded hook type -- see
Parts List.)

Jumper Panel

A 25-wire ribbon cable is cut in half; its ends are
stripped and soldered to their respective pins on the headers. The male end
of our cable (which has one of each gender) goes to the header nearest the
hinge of the box, while the female half of the cable goes to the header nearest the user.

With the male pins facing you and with the long row (13 pins) on top,
the pins are numbered: pins 1 through 13, from left to right, along the top
row, and pins 14 through 25, from left to right, along the bottom row. On the
other hand, with the socket oriented similarly, pins 1 through 13 are counted
from right to left, with pins 14 through 25 being counted from right to left
on the bottom. In both cases, the headers have their long rows away from you;
pin 1 of both is the upper left pin (treat them as male plugs). Trace these
wires with a continuity tester to make sure you know which one is which.

In addition to baring the wires from the ribbon cables, a jumper is run
from pin 1 to pin 1 of the headers, with another jumper going from pin 7 to
pin 7. Pins 1 and 7 are separate grounds.

On pins 2 and 3 between the headers is a polarity-reversing switch (a
DPDT toggle with a center-off position). Position 1 of pole A goes to position 2 of pole B; position 2 of pole A goes to position 1 of pole B. The arm
of pole A goes to pin 2 of the header which is farthest from the hinge; the
arm of pole B goes to this header's pin 3. Position 1 of A goes to pin 2 of
the other header; position 1 of B goes to its pin 3.

Accessories

For the jumper wires, little push-on sockets (which fit the wire-wrap
pins) were used -- soldered at the ends of short pieces of stranded hookup
wire. Besides single jumpers, it is sometimes necessary to make more than one
connection to a pin; for this reason, we made two or three jumpers with one of
its sockets having a wire-wrap pin riding piggy-back on it. In the latter
case, the wire-wrap pin was forced into the back of the socket, the hookup
wire was then wrapped around the assembly, and the whole mess was soldered and
covered with heat-shrinkable tubing (or wrapped with bits of tape).

Finally, it is sometimes necessary to tie a pin high or low; for this, a
9-volt battery clip is fitted with push-on terminals. The negative battery
lead goes to one well-marked push-on terminal, while the positive battery lead
goes through a 470-ohm current-limiting resistor to the other push-on terminal.

Board Layout

A 3- by 4-1/2-inch piece of perforated board is used.

Picturing it from the component side, the placement of major items is as
follows:

The Circuit Common bus is underneath the lower edge of the board. The
four transistors are at the lower right, taking up perhaps a square inch of
space. Just above these is the 566 wobbler circuit, near the top right corner
of the board. Perhaps a half-inch to the left of all this is the zener and
its string of associated resistors, stretching from top to bottom. Signal
Ground can be a short line which runs to the right along the bottom of the
board, going right between the wobbler and the transistors.

Near the zener, and mainly to its left, the various electrolytic bypass
capacitors can be placed; use your own judgment. To the left of these --
nearing the left end of the board -- put the 556 and 324, placed end to end,
reaching from bottom to top. Position these chips so that there is the better
part of an inch between them (with the 324 nearest the top edge); this way,
you will have room for all those charging resistors and diodes.

Parts List

Capacitors:

• 1--0.001uF disc
• 2--0.022uF mylar
• 2--0.1uF disc
• 2--0.22uF disc or mylar
• 1--5uF 10-volt electrolytic
• 4--100uF 10-volt electrolytic

Resistors (all but the 10-ohm unit being
1/4-watt, 5%):

• 1--10 ohms 1/2-watt 5%
• 2--47 ohms
• 3--100 ohms
• 2--220 ohms
• 2--1K
• 4--2.2K
• 2--12K
• 2--22K
• 4--47K
• 1--82K
• 2--220K
• 2--330K
• 4--560K

Semiconductors:

• 1--1N752 (or equivalent) 5.6V zener diode
• 8--1N914, 1N4148, or other small-signal
  diodes
• 1--1N4001, or other low-PIV rectifier
• 2--2N2222 (or 2N2222A) transistors
• 2--2N2907 (or any complement to the 2222)
  transistors
• 1--LM324N quad op-amp
• 1--NE556 (or LM556N, as from Jameco) dual
  timer
• 1--NE566V VCO function generator

Switches:

• 1--SPST toggle
• 1--SPDT toggle (on-off-on type) for logic-
  probe standby
• 1--SPDT toggle (on-off-on type) for data
  lines 2 and 3

Connectors:

• 2--50-pin wire-wrap headers, Jameco 923866R
• Several--Amphenol 2-20-SO2-100 sockets (for
  jumper wires)
• 1--E-Z-Hook Mini-Hook Jumper (to be cut in
  half) No. 204-l2W-Red
• 1--Ribbon cable, Jameco CDB25P-4S (male to
  female)

Miscellaneous:

• 2--small loudspeakers
• 1--open-circuit earphone jack of your choice
• 1--file-card box for 3 by 5 cards

Notes on the RS232C "Standard?"

The RS232C system is a "serial interface;" you send strings of "1's" and
"0's" down a wire, and you get strings of "1's" and "0's" back on another
wire. (A "parallel interface" would set up a row of data lines -- probably
eight or sixteen; a "hand-shaking" pin would then say "Read that!" This would
be done instantaneously, all at once, instead of "serially.") For the "serial
interface," you need three wires ("Transmit," "Receive," and "Ground"), yes?
Wrong! And the engineers begat the following:

RS232C Pin Assignments

• Pin 1--Protective Ground (the cabinets of
  the equipment)
• Pin 2--Transmitted Data
• Pin 3--Received Data
• Pin 4--Request to Send
• Pin 5--Clear to Send
• Pin 6--Data Set Ready
• Pin 7--Signal Ground
• Pin 8--Receive-Line Signal Detector
• Pin 9--Reserved for Data Set Testing
• Pin 10--Reserved for Data Set Testing
• Pin 11--Unassigned [Finally!]
• Pin 12--Secondary "Received Signal-Line
  Detector"
• Pin 13--Secondary "Clear to Send"
• Pin 14--Secondary "Transmitted Data"
• Pin 15--Transmission Signal-Element Timing
• Pin 16--Secondary "Received Data"
• Pin 17--Receiver Signal-Element Timing
• Pin 18--Unassigned [Whew!]
• Pin 19--Secondary "Request to Send"
• Pin 20--Data Terminal Ready
• Pin 21--Signal-Quality Detector
• Pin 22--Ring Indicator
• Pin 23--Data-Signal Rate Selector
• Pin 24--Transmitted Signal-Element Timing
• Pin 25--Unassigned [Just happened to have it
  lying around.]

The sins of this interface system (which was decided upon in 1969) go
back to its intended use; computer terminals and printers were all connected
via MODEM's and phone lines. ("MODEM" stands for "MOdulator/DEModulator;" no
wonder it's never been obvious. The outputs of MODEM's send Audio Frequency-
Shift Keying over the telephone.) All of these terminals and printers receive
their data on pin 3, and transmit their data on pin 2. (The above pin assignments are given as you look at the 25-pin connector of the "terminal equipment.")

Well, all of our computers fit on the desk, now, and we don't use MODEM's as the main link anymore. Therefore, my printer, which is "Data Terminal Equipment" (DTE) is run from a socket on my personal computer, which has
to pretend that it's Data Communications Equipment (DCE). (Data Communications Equipment used to refer to the MODEM.) Just as there are male and
female plugs, so there are two kinds of ports, DTE and DCE. A DCE thing
transmits on pin 3 and receives on pin 2, just the reverse of the printers and
things. A DTE thing has to be hooked to a DCE thing, or it won't work.

It got more complicated with the advent of gadgets which are so versatile that we don't know what to call them anymore. For example, with a Versa-
Braille, you can read data, write data, or store and retrieve data. What kind
of port should it have? When it "talks to a computer," it has to be DTE; when
it "talks to a printer," it has to be DCE. It was chosen to be DTE.

Just as with mismatched male and female connectors, you can make "adapter cables" which change a port from one kind to the other. Pins 2 and 3, the
data lines, must be reverse-connected in this cable, and hand-shaking pins may
have to be reversed as well. Since there is no MODEM, yet the terminal devices think (because of the adapter cable) that they are talking through

MODEM's, this kind of cable is often called a "Null MODEM" -- a MODEM that
isn't there. (In concept, this is similar to a mirage, which is where a
little man, who isn't there, keeps his car.)

It's not so bad finding out which gender of device you have, DTE or DCE. The
problems often come in when you try ciphering out the "hand-shaking" systems.

"Hand-shaking," a process of opening and closing communications, may be done
in many ways, but it comes in two classes, "hardware hand-shaking" and "software hand-shaking." Software handshaking is done via extra character strings
on the data lines, pin 2 and 3 (the characters being called "XON" and "XOFF").
Hardware hand-shaking is done by bouncing wires up and down between logic
levels "0" and "1." Back when the interface was designed, the conversation
between the "terminal equipment" and the MODEM was envisioned as something
like this:

MODEM: "Data set ready."
Terminal: "O.K., data terminal ready."
MODEM: "O.K., request to send."
Terminal: "Clear to send."

The hands would shake, and little bits of data would go back and forth
and be synchronized or verified, and the things would talk over the phone
lines faster than Aunt Clara (or Uncle Clarence). Nowadays, the data strings
are merely punctuated by the operation of one or two hand-shaking lines. Most
commonly, pin 20 is considered important by the devices; the printer stops and
starts the computer by bringing this line up and down, thus making the computer send no more data than the printer can print. (More and more printers have
buffers to soak up all the data at full speed, but the hand-shaking signals
are still used to keep everything in order.)

Suppose you have two DCE devices "talking to each other," then what?
You have to do something with the hand-shaking lines. Pin 20 (Data Terminal
Ready) may be used to operate pin 6 (Data Set Ready). The Null MODEM cable
will have to exchange these pairs of pins as well.

You seldom know exactly what connection configurations you're getting
into. Some manufacturers use pins 4 and 5 heavily. Some printers use pin 19
(instead of pin 20) as their Data Terminal Ready -- for unknown reasons. It
is for these factors, dear reader, that you need a test instrument. The
jumper system in the "breakout" box allows you to try things; an oscilloscope
or these audible "logic probes" help you discover what's on the various lines.

The logic probes in this breakout box should tell you if the "high's"
and "low's" are good (above 2.8V and below minus 2.8V, respectively). In
addition, if you think there should be an input on a particular pin, you can
find out; the input impedances are between 3 and 7 Kohms, and a good input
will not look "open."

Accessible books on the RS232C system are beginning to appear, and I
will keep you abreast of their status. In the meantime, you're on your own,
since I've expounded more than I know about the systems. Of this paper, let
it be said that, "Never has some one said so much about so many pins about
which he knows so little."

Supplier Addresses:
Amphenol Products, 2l22 York Rd.,
Oak Brook, IL 6052l
E-Z-Hook Company, 225 N. Second Ave.,
Arcadia, CA 9l006 (2l3) 446-6l75
Jameco Electronics, l355 Shoreway Rd.,
Belmont, CA 94002 415-592-8097