TABLE OF CONTENTS

THE SILICON-SYSTEMS SSI202 TOUCHTONE DECODER CHIP

FROM PAPER TO PROJECT, Part V

FINALLY, A 9-VOLT BATTERY HOLDER

THE SILICON-SYSTEMS SSI202 TOUCHTONE DECODER CHIP

Abstract

This 18-pin chip can take an audio signal containing speech and telephone "touchtones," separate them (so as not to false trigger from other audio), and present the touchtone information in binary formats. (Actually, there are two related chips, the SSI202 and the SSI203; the difference in the two chips is fairly trivial in that the SSI203 has an additional handshaking pin.) Such chips are formally called "DTMF" (dual-tone multifrequency) receivers. They're just the thing for controlling stuff by radio or over phone lines. Specific applications will be left up to you. The SSI202 is available from Radio Shack as Cat. No. 276-1303. As far as "encoders" are concerned, some very nice ready-to-go keypads are available with the touchtone generators built into them; these are available from Pipo Engineering, and specific ones of these are described at the end of this article.

Introductory Comments

As you know, the pushbuttons on touchtone phones cause pairs of tones to be generated which, when detected by equipment in the switching office, correlate with numbers that originally were "counted" as short-circuit pulses from a rotary dial. With expansion in mind, the designers of the touchtone system picked enough tones to accommodate a 16-button keypad. (On standard telephone instruments, only 12 keys are used, but an additional column to the right of these would be permissible.) On a 16button pad, the layout of symbols is as follows:

The top row of buttons includes (from left to right): "1," "2," "3," and "A." The next row down contains (from left to right): "4," "5," "6," and "B." The third row down contains: "7," "8," "9," and "C." The bottom left key is an "asterisk" (called "star" from here on). To the right of the "star" is "0." Next comes the oft' misnamed symbol, the "octothorp." The bottom-right key is "D."

Table I
Touchtone Frequencies

Note: The tones used come in two groups of four; the low-tone group pertains to the four horizontal rows of the keypad, while the four high tones pertain to the four columns of the 16-button pad. On standard telephones, the right column, which would generate 1633Hz if it existed, is not present.

• Low-Tone Group:
  • 1, 2, 3, and A--697Hz
  • 4, 5, 6, and B--770Hz
  • 7, 8, 9, and C--852Hz
  • Star, 0, Octothorp, and D--941Hz
• High-Tone Group:
  • 1, 4, 7, and Star--1209Hz
  • 2, 5, 8, and 0--1366Hz
  • 3, 6, 9, and Octothorp--1477Hz
  • A, B, C, and D--1633Hz
• Dial Tone:
  • 350Hz and 440Hz

A lot of study went into the phone company's choice of tones and other parameters. The frequency pairs which were chosen are unlikely harmonic combinations to be found in speech. Purity of the tones must be high (harmonics having to be 20dB below the fundamental frequencies) so that receiving devices can reject simple frequency "look-alikes" that might sometimes occur in speech. Timing is scrutinized; if a pair of tones doesn't "hang in there" for 40 milliseconds, it will be rejected. Even the space between tones is assessed; a 40-millisecond pause must be present in order for a succeeding pair to be accepted. All in all, it's pretty hard to fool the system with straight speech sounds.

Home brewing an equivalent decoding system, before the magic of integrated circuits, was very complicated, and was often done with less scrutiny of parameters. In the early days of automatically controlled ham-radio repeaters, your editor has been known to send many a controller on a fool's errand with a simple "Howdy dooh!" These chips have all the filters, and perform all the timing tests, so as to keep false entries from happening.

In order to simplify the circuitry following the SSI202 decoder chip, the immunity you need--from false signals--is worth debating. The SSI202 has a "Data Valid" line which, if used, would take maximum advantage of the assessment of parameter criteria. But if all you're doing is controlling a tape machine through a two-wire line which will never see any audio, then the simplest possible system will be sufficient. Then again, if using this two-wire line to carry audio means that the tape machine would turn on unexpectedly once in a blue moon, what of it; it may not be worth installing complicated circuitry to prevent this.

On the other hand, nothing can save you if you talk while the tone pair is on; everything drops out if any other audio is greater than 12dB below the weakest tone of the pair. To remedy this latter effect, you have to arrange to mute all other audio while tones are being pressed--or simply keep your mouth shut. In a word, any extra sounds "delegitimize" a tone pair, which is why touchtone telephones mute the mouthpiece when a button is pressed.

[I offer the above precautions as a responsible editor, but I never saw the decoder falsely interpret speech as a tone pair. While looking at the Data Valid Output, I hooted and barked like a man possessed, but never fooled the system.]

Technical Description

[Note: Except for pin 6, the SSI203 is the same as the SSI202. On the 202, pin 6 has no connection. On the 203, pin 6 gives the user an immediate indication as to when a tone pair has been detected, and immediately drops out when the tone pair's absence is noted. A more useful pin, pin 15 ("Data Valid"), insists that the tone pair be recognized and timed first before an "on" response and "off" response occurs. Because the SSI202 is readily available from Radio Shack, and since the simplistic signal from pin 6 of the 203 would have appeal in only special circumstances, only the SSI202 will be mentioned from here on.]

The Silicon Systems Inc. SSI202 (Radio Shack 276-1303), by using "switched-capacitance techniques," has all the filters to accomplish the tasks discussed above. (For more on "switched-capacitor filters," see "The Reticon R5620 Programmable Filter IC," SKTF, Summer 1983.) It runs its innards off a crystal-controlled time-base, and has timing circuitry to perform those needed tests as well. A sketchy block diagram follows:

Block Diagram

The input goes through a 60Hz reject filter, then through a preamplifier. This preamp goes into a pair of filters--called "band-split filters"--where the high-frequency "column tones" are separated from the low-frequency tones of the rows.

Further on, band-pass filters separate individual tones from their group, whereupon they are subjected to amplitude testing--there being a reference voltage to compare them with. The information about the four frequencies of each band then goes to a decoder which has the four tristate outputs we have access to. Meanwhile, information from the amplitude detectors is subjected to timing tests that orchestrate the decoder's presentation of data and validity signals to the chip pins. (Actually, the final output section, which is manipulated by the timing circuitry, is probably a parallel-in parallel-out shift register.)

Running this whole mess is a crystal oscillator (made from an inverter), whose output frequency is then divided by eight. The crystal frequency is about 3.58MHz, so that the time-base which runs the chip is 447Khz.

Specifications

• Power Supply Voltage--typically 5V, maximum 7V.
• Power Supply Current--minimum 10mA, maximum 16mA.
• Tolerance to Power Supply Noise--must be below 10mV peak-to-peak.
• Input Impedance--from plus V down to a value of V minus 10V, 100K shunted by 15pF.
• Tolerance to Tone Frequencies--at minimum plus or minus 1% plus 2Hz, typically plus or minus 2.3% plus 2Hz, with a maximum tolerance of plus or minus 3.5% plus 2Hz.
• Acceptable Tone Amplitudes--from minus 32dBm to minus 2dBm (referenced to 600 ohms).
• Input Noise Tolerance--12dB below weakest touchtone.
• Purity of Touchtones--harmonics must be 20dB below fundamental.
• Acceptable "Twist" (ratio of tone amplitudes)--tones must be within 10dB of each other.
• Tolerance to Dial Tone Amplitude--The amplitude of the dial tone can be equal to that of the weakest touchtone, but no louder, or detection of touchtones may not occur.
• Digital Output Voltages--for logic "0," minimum 0V, maximum 0.5V (with 400uA load to plus V); for logic "1," maximum plus V, minimum V minus 0.5V, (with 200uA load from output to ground).
• Digital Input Tolerances--for logic "0," must be below 30% of plus V; for logic "1," must be greater than 70% of plus V.
• Minimum Data Clear Pulse Width--200 nanoseconds.

The input signal need not be capacitively coupled to pin 9 unless there is a chance of exceeding the voltage levels stated above. The upper limit is the positive supply, while the lower limit is 10 volts below this (minus 5 volts, if a 5-volt supply is used). The coupling capacitor recommended is 0.01uF.

The SSI202 contains an inverter intended for use with a "color-burst" crystal (Radio Shack 276-1310, 3.579545MHz). This crystal, shunted by a resistor, is placed between pins 11 and 12 (pin 11 is this inverter's output, while pin 12 is its input).

[The literature is rather inconsistent when specifying this feedback resistor value. The Silicon Systems data sheet specifies 1 megohm. In the text of the Radio Shack material, 1 megohm is also stated. However, in all the sample circuits shown in the Radio Shack pamphlet, this resistor is often shown to be 10 megohms. Perhaps this suggests that, at one time or other, someone in a laboratory had trouble getting an oscillator to take off--solving this problem by reducing the negative feedback across the inverter. With the editor's single sample, everything worked fine with either value--1 megohm or 10 megohms. Therefore, 1 megohm will be shown shunting the crystal here; but if you have trouble, try increasing this value.]

Up to ten devices can be run off of one whose crystal oscillator is enabled. To do this, the "Auxiliary Time Base Output" (pin 13) of the first is connected to that same pin (pin 13) of the others. Furthermore, each additional 202 running off pin 13 of the master should have its pin 8 grounded (disabling the oscillator section), and should have pin 12 tied high (thus preventing the "input" of the oscillator's inverter from floating).

This time-base output, pin 13, operates at a frequency of one-eighth that of the crystal frequency, 447.44kHz. Although pin 11 is an output which drives the crystal, it is intended to drive reactive loads, and should not be used for other purposes.

The literature states that, if the crystal frequency is needed for other circuitry, a separate gate should be used as the crystal oscillator, with a buffered version of this feeding into pin 12 of the SSI202 chips. If pin 12 is driven by an external 3.58MHz signal, this section of the chip must be enabled by tying pin 8 high.

The binary outputs are "push-pull" (made with complementary pairs), so that they can pull current from either ground or plus V. These outputs are sufficient to drive low-power TTL devices. What's more, they can be thrown into the tristate condition by bringing pin 3 low. The "Data Valid" line is an exception; it cannot be made tristate.

All the outputs rest at zero until tones are detected. When any tone pair is considered valid, the Data Valid pin goes high (within 46 milliseconds). Shortly before this (a minimum of 7 microseconds), the appropriate "Data" lines will have gone high. After the tone pair ceases, the data will be left there for between 35 and 50 milliseconds, with the Data Valid line dropping out at least 4.2 milliseconds before this. The point is, data will be there before and after the validity signal appears.

In a handshaking scheme, the Data Valid pin may be "polled" every 40 milliseconds to see if a new valid tone pair exists, whereupon it can be "cleared" in preparation for detecting the next one. However, this scheme gets rid of "valid pause" information; if you want this back, the data outputs must be monitored with a 4-input NOR gate to note when they fall away. (This may explain why "0" is not 0000 in the hexadecimal setup; instead, the rare entry "D" is 0000, which would fool this kind of handshaking system.)

This brings to mind the SSI203, which has an additional handshaking output on pin 6 (pin 6 has no connection on the 202). On the 203, pin 6 is called an "Early Detect Output"; it goes high as soon as a tone pair is detected within the chip (taking no longer than 22 milliseconds), and drops low as soon as detection ceases (in no more than 18 milliseconds). In other words, the data on the binary output lines are guaranteed to be valid during the time that Data Valid is high, but there is no such guarantee for the Early Detect line. In asking what good this new pin might be, it occurred to me that the Early Detect line would not be operated by the Data Valid's "Clear" terminal, which could simplify a circuit where the flawless detection of number "D" is to be done.

Table II
The So-Called Hexadecimal Output Listing

Notes: This mode is gotten by tying pin 2 high. The numbers "1" through "9" are how you would expect--"1" being 0001, and "9" being 1001. However, the touchtone "0" is not 0000, but is 1010 (the hexadecimal number A). The 16-button keypad position "A" is not the hexadecimal A, but is hex number D. "D" on the keypad gives you a binary 0000. Just follow the table without preconceived notions.

• "1"--0001
• "2"--0010
• "3"--0011
• "4"--0100
• "5"--0101
• "6"--0110
• "7"--0111
• "8"--1000
• "9"--1001
• "0"--1010
• "Star"--1011
• "Octothorp"--1100
• "A"--1101
• "B"--1110
• "C"--1111
• "D"--0000

Table III
The So-Called Binary-Coded 2-of-8 Output Listing

Notes: This output code is "spatial" in nature with respect to a 16-button touchtone pad. If you forget about the symbols assigned to the keys, you will notice that the binary code increases consecutively as you go from left to right: on the top row, then on the row beneath, etc. It's no accident that this is the way the tones progress; every time you jump down a row, the lower tone of the pair goes up a notch. Actually, this output arrangement makes it convenient for the digitally logical mastermind to get a row of outputs which correlate to frequency--and hence physical keypad layout.

If you want this arrangement, pin 2 ("Hex/B28") will be grounded. In addition, pin 4 ("IN1633") must also be tied low; if not, the decoder will consider the whole right column of the 16-button pad to be invalid.

The term "2-of-8" refers to a hookup shown in the literature which permits display of the two tones on eight outputs. This is very fine, I suppose, but one's practical use of the chip can be much simpler. If the "1-of-16" decoder described later is used, you get sixteen individual outputs, in order, as you progress through the pad. This seems of more use to your editor, and I'll leave it at that.

• "1"--0000
• "2"--0001
• "3"--0010
• "A"--0011
• "4"--0100
• "5"--0101
• "6"--0110
• "B"--0111
• "7"--1000
• "8"--1001
• "9"--1010
• "C"--1011
• "Star"--1100
• "0"--1101
• "Octothorp"--1110
• "D"--1111

Discussion of Circuits

Input Considerations

A lot of fuss is made about interfacing this chip directly with the phone line, only to end the story with the crushing realization that you would have to register your setup with the Federal Communications Commission. An application for "type acceptance" would have to be approved. Because their circuits are educational, a summarized version will be given here. However, no automatic answering system was included, so the circuits given--by themselves--would be of limited application.

You can avoid the legal matter and solve your automatic answering needs by stealing an audio signal from the speaker in your answering machine; the cold side of the speaker will go to "circuit common" of the decoder, with the other side of the speaker going through 0.01uF to pin 9 of the SSI202.

Suggested Connection to the Phone Line

In the Radio Shack booklet that comes with the 276-1303, the following basic arrangement was shown: First going through a "hook switch," the two wires of the phone line go to the primary winding of a 1-to-1 600-ohm line transformer. Shunting this primary is a 150-volt metal-oxide varistor (MOV) to suppress high-voltage transients. Across the secondary are series-connected zener diodes (39-volt units connected back-to-back). One side of this secondary is grounded, while the top end goes through a suitable diode clamping arrangement to pin 9 of the SSI202.

While the phone company standards state that the touchtone levels can be from minus 24dBm to plus 6dBm, the chip puts an upper limit of minus 2dBm on you (about 0.5 volts RMS). (These levels are based on a line impedance of 600 ohms.) An 8-ohm speaker running at 30 milliwatts will be at this level; a low listening volume. (I didn't damage the chip by turning things up, but I expect "falsing" would be the result of too high a level.)

Where necessary (if the DC level on the source exceeds the chip supply), they recommend putting 0.01uF in series with the hot input lead, pin 9. The input impedance is 100K; this is so only if you stay within a range of 10 volts below the plus supply line. (Internal clamping of some sort confines you to this zone.) Since my ham-radio receiver runs off 12 volts, and since it has a large-value coupling capacitor to its earphone jack, I have included this capacitor in these sample circuits.

The 12dB signal-to-noise ratio they insist upon is only good for wide-band noise; usually speech or other program material has most of the power concentrated in the lower portion of the audio spectrum. However, if there is noise specific to the band between 1kHz and 6kHz, this must be kept 20dB below the weakest tone to be detected.

A surprising source of noise may result if the line contains noise above 28kHz. This comes from the fact that high-frequency noise can be "mixed" with a 56kHz internal sampling frequency (used for the switched-capacitor filters)--the end product winding up in the audio spectrum. For this kind of noise, they recommend installing a 6.6kHz filter as follows:

A 6.6kHz Noise Filter

Pin 9 is shunted to ground by 0.01uF; pin 9 also goes through 2.4K to the source. This will be necessary only for projects containing digital events.

Sometimes the purity of the touchtones is in question. For these cases, they recommend a 3.1kHz filter as follows:

A 3.1kHz Filter for Touchtone Harmonic Suppression

Pin 9 is shunted to ground by 0.01uF. Pin 9 also goes through 5.1K to the source.

Test Circuit

Pins 7 and 10 are grounded, as is the negative side of the 5-volt supply. Pin 5 goes to plus 5 volts. Pin 5 is bypassed to ground by 0.05uF. Both enable pins, pins 3 and 8, are tied high. Pin 6 has no connection. The Data Valid Clear pin, pin 15, is grounded.

The Hex/B28 pin, pin 2, can be put where you like (see text). I tied it high. You also have your choice as to where pin 4 should go; if high, the chip will ignore the right-hand column of a 16-button touchtone pad. I grounded pin 4 so as to preserve the fourth-column option.

The color-burst crystal (276-1310) goes between pins 11 and 12. Also between 11 and 12 is a 1 megohm resistor.

Pin 9 goes through 0.01uF to the earphone output of a radio receiver (or some such source). Or, this source can be the audio lead from one of the Pipo touchtone encoders listed at the end of this article. (The ground lead of the Pipo device should be grounded to this circuit as well.)

Finally, a voltmeter or logic probe is used to look at pins 1, 18, 17, and 16. (Pin 1 is the least significant bit, while pin 16 is the most significant.)

Sixteen unique outputs, one for each tone, may be had by "decoding" the outputs of the SSI202 with a CD4514 1-of-16 decoder. This is done as follows:

Deriving Sixteen Unique Outputs

Pins 1, 18, 17, and 16 go to the 4514's pins 2, 3, 21, and 22, respectively. The Data Valid, pin 14, goes to the "strobe input," pin 1, of the 4514. On the 4514, pin 12 is grounded; pin 24 goes to plus V. Pin 23 of the 4514 ("NOT Enable") is grounded.

On the 4514, the sixteen output pins representing 0000 through 1111 are as follows: 11, 9, 10, 8, 7, 6, 5, 4, 18, 17, 20, 19, 14, 13, 16, 15. Now, as if that ain't bad enough, remember that in the hexadecimal setup (pin 2 of the SSI202 tied high), its outputs are not quite in order; that is, it leads off at 0001 and doesn't even have a zero until you get to entry "D" on a 16-button keypad. The following table correlates the hexadecimal-decoded 12-button keypad with the outputs of the 4514.

• "1"--Pin 9
• "2"--Pin 10
• "3"--Pin 8
• "4"--Pin 7
• "5"--Pin 6
• "6"--Pin 5
• "7"--Pin 4
• "8"--Pin 18
• "9"--Pin 17
• "0"--Pin 20
• Star--Pin 19
• Octothorp--Pin 14

Multiple-Entry Decoding Scheme

Although we have never actually built one, Tom Fowle suggested the logic scheme with which a number of two or more digits could be required to control something. An exemplary application might be: Suppose a ham-radio operator wants to have a way by which he/she can alert the family to an imminent arrival home. It would be unfair to make them monitor the radio all day waiting for W6PDQ to announce himself; it would be much more civilized to mute the receiver until a 2-digit number is transmitted by the old goat. The decoder system would be looking for those two specific digits, and when it "heard" them, it would close a relay and connect the loudspeaker.

The answer is to set up a string of R-S flip-flops (one for each digit); each is "set" by an output of the 1-of-16 decoder (4514, see above). Suppose the secret code selected is the number 25. These would come up on the CD4514 as pins 10 and 6, respectively.

The "Set" terminal of the first flip-flop goes to pin 10 of the 4514 (the entry "2" will set this flip-flop, bringing its "Q" output high). Next, a 2-input AND gate is set up to sense the coincidence of the next desired number and the logic high of this flip-flop. In other words, one input of the AND gate goes to the "Q" output of the first flip-flop, with the other gate input going to the output pin of the next number (pin 6 of the 4514). The output of this AND gate goes to the "Set" terminal of a second flip-flop (an entry "5" must follow the entry "2" in order to "set" this one). Each additional digit requires a 2-input AND gate with one input on the next digit pin of the 4514, and the other input looking at the "Q" output of the previous flip-flop.

But what's to keep W6PDQ from entering a "2" in the morning, and then a "5" later on? The answer is to give him a time limit, say two seconds, which refreshes at each entry. This could be done with a CD4528 (see SKTF, Summer 1981)--a dual retriggerable one-shot. The first one-shot would be set for two seconds, and would be triggered from the Data Valid pin of the SSI decoder. The output of this first one-shot would then trigger the second--set for a very short time, perhaps 10 milliseconds--with the latter used to "reset" the R-S flip-flops. (Therefore, the "Resets" are tied together and go to the output of the second one-shot.)

Humm-mm-mm, now what's to keep unauthorized personnel from just punching one number after another--cycling through the touchpad, then going in reverse, etc.--until the desired numbers come up in sequence. (Thus far, as long as he pushes them faster than every second and a half or so, the circuit will just patiently wait for the right sequence to come by). The answer here is to put an AND gate on some unused numbers, and use the output of this (AND'ed with the second one-shot) to reset the flip-flops. The more unused numbers you choose to detect to abort the entry, the more secure will be the system.

Now, before you order parts for our new invention, it's only fair to tell you that this very device is commonly available at ham stores--for $70 or so.

Pin Connections
SSI202 and SSI203
Touchtone Decoder Chips

[Note: Of all things, the definition of the "Ground" pin is unclear in the literature. Both pins 7 and 10 are listed as "Ground" or "minus V." Funny, though; with a continuity tester, they don't look connected at all. With the circuit running, ungrounding pin 7 seems to make no difference. On the other hand, with pin 7 grounded, ungrounding pin 10 causes the circuit to float up to plus V (and stop working); pin 10 is "Serious Ground." In one of the Radio Shack circuits, they called pin 7 "V sub NA," whatever that means. Anyhow, all circuit samples show both 7 and 10 grounded, so we'll follow directions.]

• Pins 7 and 10--Ground (all circuits show both of these pins wired to ground.)
• Pin 5--Plus V
• Pins 11 and 12--Shunted by the crystal in parallel with 1 megohm.
• Pin 8--Crystal Oscillator Enable (high to enable)
• Pin 13--Auxiliary Time Base Output (1/8 of the crystal frequency)
• Pin 4--IN1633 (If low, permits input of 1633Hz tone; if high, this frequency will be rejected. This 1633Hz is present in the fourth column of a 16-button touchtone pad.
• Pin 9--Analog IN
• Pin 16--Output D8 (most significant bit)
• Pin 17--Output D4
• Pin 18--Output D2
• Pin 1--Output D1 (least significant bit) Pin 3--Output Enable (high to enable)
• Pin 2--Hex/B28 (Hexadecimal or Binary "2 of 8")
• Pin 14--Data Valid Output; goes high when a valid tone pair is detected. It goes low with detection of a "valid pause"--or when its clear terminal is brought high--whichever comes first.
• Pin 15--Clear Data Valid (high for clear)
• Pin 6--No connection for the SSI202; so-called Early Detect for the SSI203.

Pin Connections
CD4514 1-of-16 Decoder

• Pin 12--VSS
• Pin 24--VDD
• Pin 23--NOT Enable (low for enable)
• Pin 1--Strobe (high to set)
• Pin 22--IN8 (most significant bit)
• Pin 21--IN4
• Pin 3--IN2
• Pin 2--IN1 (least significant bit)
• Pin 11--OUT0
• Pin 9--OUT1
• Pin 10--OUT2
• Pin 8--OUT3
• Pin 7--OUT4
• Pin 6--OUT5
• Pin 5--OUT6
• Pin 4--OUT7
• Pin 18--OUT8
• Pin 17--OUT9
• Pin 20--OUT10
• Pin 19--OUT11
• Pin 14--OUT12
• Pin 13--OUT13
• Pin 16--OUT14
• Pin 15--OUT15

The Pipo Engineering Encoders

The editor chose these encoders because, although they are rather costly (in the $60 class), they save a lot of trouble. Besides their keypads, their circuitry is all self-contained. [One could buy a keypad and the matching encoder chip (Motorola MC14410P), but that would lead to another article like this one.] The Pipo units are all miniscule in size (compared to what I could build), they present you with as few as three hookup leads, and are easy to mount.

The specifications given here are from the Pipo catalogue, not from actual data sheets. Some of the "specs"--like current drain--are not quite clear to me, but I'll forge ahead as follows:

There are various added features which you can get: Fancy ones are illuminated. Others have "sounders" that "pip" as you press the keys. The fanciest have both illumination and sounders. All can be gotten with a "push-to-transmit" relay whose "hang time" is adjustable (after a button is pressed, the relay will stay closed for 0 to 3 seconds, this duration being adjustable via a trim pot on the back).

All units have an audio-level adjustment accessible from the front (a screwdriver-adjustable multi-turn trim pot). They say that the maximum output is plus 2dB (3V peak-to-peak on a 600-ohm line); since the decoders in this article only need from minus 32dBm to minus 2dBm, there is plenty of level to play with.

The most basic units have "three-wire hookup"--battery positive, battery negative, and audio out. They are "polarity protected," so you can't damage one by hooking it up backwards. (However, they don't say that there is a capacitor in the audio line, so putting DC on this line may be hazardous.) It should be noted that the lower limits of supply voltage requirements vary with the features selected. Basic units can operate on voltages from 6 to 16 volts. On these basic units, the maximum current drain--with a button depressed--is 20mA. Apparently, there are voltage regulators in them which take over at 9.5 or 10 volts. The specs suggest this by stating that below this level, the idling current is 1 microamp, but above this voltage, the idling current is 1 milliamp. On units with a sounder (such as the PP-4), the operating voltage is from 9 to 16 volts, with idling and "key-down" currents of 6mA and 18mA, respectively. Those with PTT relays require operating voltages of 8.5 to 16 volts; their idling and operating currents are 1mA and 20 mA, respectively.

They tell you that these things actually have relays in them. This seems hard to believe, since the current drain does not reflect this extravagance; I would have expected an FET switch. Whatever it is, the "contacts" are good for 110mA at 28V DC. The relay closes as soon as a key is depressed. As obtained from the factory, the hang-time (opening delay time) is set for 2 seconds. This delay can be adjusted from 0 to 3 seconds via a trim pot on the back.

Besides its intended use with radio transmitters, the above relay could be used to mute unwanted audio while numbers are being entered. This would prevent "confusing" of the SSI decoding chip with simultaneous tones and speech, which will ruin your entry. This relay closes instead of opening, so it will be left to the brilliant reader to devise a way of shorting out his/her unwanted audio.

• PP-1--Most basic 12-key pad; operates on 6 to 16 volts. Price: $55. PP-1K--12-key pad with PTT relay; operates from 8.5 to 16 volts. Price: $62.
• PP-2--Most basic 16-key pad; operates from 6 to 16 volts. Price: $59. PP-2K--16-key pad with PTT relay; operates from 8.5 to 16 volts. Price: $66.
• PP-4--Same as PP-1 (12-key) with added "sounder" for user feedback; operates from 9 to 16 volts. Price: $55.
• PP-4K--Same as PP-4 (12-key) with PTT relay; operates from 9 to 16 volts. Price: $62.

Pipo Communications
P.O. Box 2020
Pollock Pines, CA 95726
(916) 644-5444

FROM PAPER TO PROJECT Part V

Abstract

A voltmeter and a milliammeter will be described. The design embodies a healthy combination of familiar IC's, variations on their circuits, and (by installing two IC's in one socket) presents the builder with a 14-pin socket. This is a grown-up project; coping with two lines of seven pins is not much different from soldering on rows of eight, nine, or more. In addition, the resulting battery of instruments--a continuity tester, a test amplifier, a test oscillator, resistance and capacitance bridges, ending with meters for voltage and current--comprises a well-equipped laboratory.

[This specific voltmeter was first designed and built by Mr. Vispi Mirza, a guest at Smith-Kettlewell in 1984. Ever since, our lab copy of it has donned the moniker "the Vispimeter" for his name's sake. The accompanying milliammeter is a trivial variation of the Vispimeter. Thank you, Vispi.]

[Note: The basic chopper-comparator portion of these instruments also makes a dynamite "meter reader" that can be used to "read" visual instruments. Specific instructions for doing this can be found in "Basic Analog Meter Reader," SKTF, Spring 1981. Furthermore, whether lazy or enterprising, there will be those among you who will choose to build a single device, the input networks being selected by a switch. As for this editor, I prefer separate instruments, since this empowers me to make simultaneous measurements. Nevertheless, voltmeters and milliammeters are traditionally combined, and instructions for doing so will be briefly outlined for the advanced builder in a section called "Embellishments and Improvements."]

Theory of Operation

The instruments described in this installment are of the so-called "null type." Like with the bridges of Part IV, readings are taken by adjusting a pointer knob--fitted with a Braille scale--until the point of "no sound" is located, whereupon the position of the pointer is read. Each meter has only two controls, the Braille-calibrated pointer knob and a range selection switch. The "range switch" also doubles as the on-off switch. The "off" position is in the clockwise direction so that, as you turn the meter on, the first range you select is the highest one; this guards against presenting the meter with a high-value unknown while in a more sensitive position, which could burn it out.

Although operation is familiar, the theory behind these meters is somewhat different from that of the bridges. (In the bridges, the unknown element is "driven" by an audio oscillator; its impedance was then balanced against a variable resistance. With a meter, you don't want to "drive" a test circuit to see how it responds, you want to "sense" it with as little invasion as possible.)

Both the voltmeter and milliammeter are actually voltage-measuring devices. The heart of each meter is a 100-millivolt meter; only the input resistor networks differ. The voltmeter works by selecting appropriate "voltage multipliers" (attenuators, as you might say) so that whatever range you wish--0 to 100 millivolts, 0 to 1 volt, 0 to 10 volts, or 0 to 100 volts--a maximum of 100 millivolts will be presented to the internal workings. (The so-called "sensitivity" of the voltmeter is 100,000 ohms per volt.) The milliammeter works by detecting the voltage across low-value resistances, the selected one of which appears across its input. Available current ranges are: 0 to 100 microamps, 0 to 1 milliamp, 0 to 10 millamps, 0 to 100 milliamps, and 0 to 1 amp. (The sensitivity of the milliammeter is 10 mhos.)

The system used in these meters is a comparison system. The meter circuitry contains a voltage standard--made with a 6.2V zener diode (see "Know Your Parts"). The 6.2 volt standard goes through a "calibration rheostat" which drops the voltage to the Braille-calibrated control to a precise 100 millivolts. Thus, we start with an adjustable standard--0 to 100 millivolts--which can "replicate" any unknown input voltage within this range. A reading is gotten by "matching" the voltages.

The rest of the circuitry comprises a way by which we can know when the voltages are matched. We will do this "matching" by "listening" to the difference with the audio amplifier; when there isn't a difference, there won't be any sound. The two voltages--the unknown and the adjustable standard--are loosely tied together via a "sampling resistor." Then, an audio amplifier is presented with this resistor's voltage drop. (Actually, unless they are "pulsating," DC voltages don't make any sound by themselves through an amplifier.) The next trick is to sample the difference voltage at an audible rate--modulating it so we can hear it.

This is accomplished by "chopping" the "difference signal" at an audible rate--then listening to it. The "chopper" is simply an electronic switch in series with the sampling resistor. This switch is then turned on and off at an audible rate. If there is any "difference voltage," an audibly interrupted version of it will be heard. As this difference is minimized by adjusting the Braille-calibrated standard, the chopped signal fades away.

The "switch" of the chopper is a photo field-effect transistor (photo FET); when light hits this little channel of semiconductor, charge carriers are knocked around and the channel becomes conductive. The G.E. H11F1 (a so-called "opto isolator") is an encapsulated pair of elements--the desired photo FET and a light-emitting diode (LED) to illuminate it. The LED is driven by a 555 oscillator, causing the photo FET, which is in series with a 47K sampling resistor, to chop the difference signal. An LM386 amplifier presents the audible signal to the user via a loudspeaker.

[Note: The opto isolator appearing in the original article was the H11F3, which is the lowest grade of three related devices; in 1981, that was cheaper (about $1.50). Since then, the H11F1, originally over $3.00, has become more available, and its price is now under $1.50. Quality is not an issue here; easy availability is.]

Know Your Parts

[Note: After "Step-by-Step Construction," there is a section for advanced students called "Embellishments and Improvements." For example, on the lowest ranges of the voltmeter, the input resistance can actually be made infinite (with very complicated switching being necessary). In addition, suggestions will be given as to how both the voltmeter and milliammeter can be built using a common "null-type" chopper-comparator circuit (thus making one instrument serve as both). If this is done, another switch must be obtained: a DPDT (double-pole double-throw) toggle, with a "center off" position, will be used for off/on and to select the mode. Note also, since it is assumed that two separate instruments will be built, the quantity of each part common to both units has been doubled in the Parts List. The only parts not common to both are the precision resistors comprising the input networks.]

Parts Common to Both Meters

Capacitors

• 4--0.01uF disc ceramic
• 4--0.1uF disc ceramic
• 2--10uF 10-volt electrolytic
• 2--22uF 10-volt electrolytic
• 2--100uF 10-volt electrolytic
• 2--220uF 10-volt electrolytic

Resistors (1/4-watt 5%, unless otherwise specified):

• 2--10 ohm 1/2-watt 5%
• 2--470 ohm
• 2--1K
• 2--2.2K
• 2--47K
• 4--100K

Potentiometers, Switches and Paraphernalia:

• 2--1meg 15- or 20-turn PC-mount trim pots, Jameco 43P-1M, Mouser 32NA601
• 2--double-pole 6-position rotary switches (a 5-position one would do for the voltmeter, but the 6-position units listed here are easy to come by), Jameco MRS260, Mouser 10YX026
• 2--10K precision linear pots, Clarostat 58C1-10K
• 2--long pointer knobs suitable for the Braille dials, G.C. 37-582 2--short pointer knobs for "range" selection, Mouser 45KN012
• 2--Braille dials (included)

Semiconductors:

• 2--6.2V zener diodes, 1N753 (available from Jameco under that number), or Radio Shack 276-561
• 2--555's, Jameco LM555N, Radio Shack 276-1723
• 2--LM386N (available from Jameco under that number), Radio Shack 276-1731
• 2--G.E. H11F3 or H11F1 opto isolators, Digi-Key H11F1GE
• 1--1N4003 (needed only for the milliammeter)

Miscellaneous:

• 2--8-pin IC sockets, Radio Shack 276-1995
• 2--14-pin IC sockets, Radio Shack 276-1999
• Some Vector Board with holes on 1/10-inch spacing, such as Vector No. 169P44ELDP
• 4--binding posts (two red and two black), Radio Shack 274-662
• 2--small speakers, Mouser 25SP016, Radio Shack 40-206
• Hatful--9-volt battery clips, Mouser 12BC095
• 2--plastic cabinets, 7-3/4 by 4-1/4 by 2-3/8, Radio Shack 270-232 Hardware--such as 1/4-inch spacers, Kulka-Smith 4006 or equivalent.

Precision Resistors

You are lucky if you have a surplus of precision 1% resistors from the "standards" of the resistance bridge; those same values will be required here. (In fact, the world's supply of 1% 10-megohm units has seemingly dried up completely, and I've had to resort to recommending a 5% unit for this value this time.) Three new values are required, although all are restated here.

Because there is one critical component (the H11F1 opto isolator) for which Digi-Key is a convenient supplier, the Digi-Key resistor part numbers are listed here wherever possible; this should ease the burden of the "minimum order" just a little. Note that, whether from Mouser or from Digi-Key, the minimum order for each resistor value is a quantity of five. Oh well, you'll use the surplus elsewhere.

• Precision Input Resistors for the Voltmeter (of any wattage, 1% where possible):
  • 1--11K, Digi-Key 11.0KX, Mouser 29MF250-11.0K
  • 1--100K, Digi-Key 100KX, Mouser 29MF250-100K
  • 1--1meg, Digi-Key 1.00MX, Mouser 29MF250-1.00M
  • 1--10meg, Mouser 29SJ250-10M is the 5% unit; this would be available in any parts store with 5% tolerance.
• Precision Shunt Resistors for the Milliammeter (1/4-watt, 1% units would be preferred, although low-value units listed here could only be found with 5% tolerance):
  • 1--0.1 ohm, Mouser ME285-KNA412-.1
  • 1--1 ohm, Digi-Key 1.00H, Mouser ME276-1.00; being 5% units listed here, any parts store may have this.
  • 1--10 ohms, Digi-Key 10.0H, Mouser ME276-10.0; being a 5% unit of common value, any parts store will have this.
  • 1--100 ohms, Digi-Key 100X, Mouser 29MF250-100 1--1K, Digi-Key 1.00KX, Mouser 29MF250-1.00K

Experimenting with the Parts

Although you could sort them by value, mixing up all the resistors now becomes risky; precision and non-precision units are starting to feel frighteningly similar. Within these two categories, though, you have the power to sort them by various methods:

Units like the 1/2-watt 10-ohm resistor of the chopper-comparator circuit, and the 0.1-ohm ammeter shunt, are different in size (the shunt found in Mouser is a 1-watt unit). Except for the low-value milliammeter shunts and the 10-ohm unit mentioned, all the other resistors will make your continuity tester generate distinctly different pitches (11K and 10K values of the precision units will be close, but different enough). Except for the 0.1 ohm and 1 ohm shunts--which after all, are different sizes-your resistance bridge can "read" all of these.

If you didn't build the resistance bridge, there is still a way of telling the 1-ohm shunt from its 10-ohm neighbor. First, find the 1-ohm and 10-ohm units with your continuity tester (they will both sound like a short). Plug a pair of test leads into a radio whose output is capable of driving an external speaker. ("Walkman"-type radios are sometimes not good enough, because they're high-impedance. A touch of irony is that you cannot use your test oscillator, either, for this test because of its 100-ohm output resistance.) Attach one clip to one side of a little speaker (such as one in this parts collection). Next place one end of both resistors in the free test clip. Alternately touch one, then the other, of the free ends to the unused speaker terminal. The 10-ohm resistor will attenuate the signal by about 3dB; this isn't much, but it's discernible.

The various capacitors will probably be sortable by size (the only possible confusion being between the 0.01uF and 0.1uF units. The electrolytic capacitors will not only be different in size, but their "charging time," when presented to the continuity tester, will be different--the larger the capacitance, the slower will be the descent of pitch. (Don't forget how to tell polarity of these electrolytics--see "Know Your Parts" of Part III, Fall 1985.) The ideal way of sorting the small units is with the capacitance bridge of Part IV. Still, with the small units (0.1uF and 0.01uF), there is a trick by which you can identify them without a bridge. [There's more than one way to skin a cap.]

Plug a set of test leads into your test amplifier; clip these across a 100-ohm resistor. Plug a second set of test leads into the test oscillator. One by one, try your disc ceramic capacitors by connecting them between the hot oscillator output lead and the hot amplifier input lead (with the 100-ohm input shunt kept in place). There will be a marked difference in loudness of the tone heard with these contrasting values; besides this, the little 0.01uF units will make the tone primarily rich in high frequencies, and the 0.1uF units will cause loading of the oscillator that will lower its pitch.

Testing the Zener Diode

The polarity of this component is important. In one direction, it acts almost like a short--0.7 volts will exist across it, just like any forward-biased diode. (Except in clever circuit designs, this "forward direction" is usually not used.) In the other direction, it looks like an open circuit until 6.2 volts is put across it, at which point it tries to maintain this voltage (sacrificing itself if there isn't the necessary "current-limiting resistor" in series with it). In this application, driven by a 9-volt battery through a 470-ohm current-limiting resistor, the zener serves as a voltage regulator, maintaining a constant voltage which is independent of battery condition (unless the battery sags below 6.2 volts).

If this zener diode is put across your continuity tester in the forward direction, the pitch will be slightly lower than with the tester's leads shorted together. In the reverse direction--which is what we actually want--a much lower tone will be heard (unless the battery in your tester is a dead one, whereupon the diode will not conduct at all). When this lower tone is present, the end going to the negative tester lead is the one which is supposed to go to ground (the anode, as stated in the circuit).

The Opto Isolator

The H11F1 (or its lower-grade brother, H11F3) is a so-called "opto isolator." As mentioned in the above theoretical section, it contains an LED and a photo FET. The circuit "switched," or otherwise operated, by the FET is "isolated optically" by the light-emitting diode that drives the FET--a handy gadget for many applications where voltages are incompatible, etc.

One might think the device would have four leads, two for the LED and two for the channel of the FET. (An electrical FET gate is not provided.) For whatever manufacturing reason, the device comes in a 6-pin dual in-line package. Like all "DIP" assemblies, the pins are numbered as counted in a clockwise direction.

With the pins facing you, and with pin 1 at the upper right, the right-hand column of pins is 1, 2, and 3 (from top to bottom); the left column is (from bottom to top) 4, 5, and 6. Pin 1 is marked by a small dent in the top of the package at the "pin 1 corner." A Braille stylus can be used to find this dent. On the device, pins 1 and 2 are for the LED, with pin 1 being the anode; pins 4 and 6 are for the FET's channel, the drain and source being interchangeable.

Coincidentally, there aren't any 6-pin sockets. The most traditional tactic is to waste an 8-pin socket on the H11F1, hoping that someone won't plug it into the wrong six holes during a repair. Another option is to choose a longer socket for one of the other 8-pin chips and let this oddball device become its roommate (8 plus 6 adds up to require a 14-pin socket). I have chosen to do this mainly so that the student will finally get to work on a 14-pin socket, which will force some planning to be done in the sequence of soldering.

Presentation of Circuit
Irrespective of Layout

This circuit is a neat combination of things: some of these you have seen before (with slight changes), plus new items (the H11F1 and the zener circuit). It is this author's hope that your experience will serve you well in conceiving of places to put things. As you read this circuit, picture that long-skinny board we are at home with and try to make portions of this circuit take shape in your mind.

On the other hand, as mentioned before, a circuit is presented so as to make "circuit sense." The order of steps you take in building something will rarely be the same as the order of topics in the circuit description. [The circuit has not been written with the idea of giving you any hints as to what to do first. In fact, except for circuit changes specific to these meters, the circuit description of the chopper-comparator section is a direct quote from an article published in SKTF, Spring 1981, called "Basic Analog Meter Reader." This is real stuff now; if you can cope with it, you're a graduate.]

Chopper-Comparator System

The negative side of the 9-volt battery is grounded. One pole of the double-pole rotary range switch is used for the on-off function. For the voltmeter, positions 1 through 4 are tied together and go to the positive of the battery. For the current meter, positions 1 through 5 are jumpered together and go to battery positive. The arm of this switch section goes to the VCC line.

The bottom of the Braille-calibrated precision pot (10K) is grounded. The top of this pot goes through a 1meg calibration rheostat to the cathode of a 6.2V zener diode (1N753). The anode of this diode is grounded. The cathode also goes through 470 ohms to the VCC line.

The arm of the 10K calibrated pot goes through 47K, then through the channel of the FET, to the "hot" output point of the front-end resistance network. (Pins 4 and 6 on the H11F3 or H11F1 are the ends of the FET's channel.)

The junction of the 47K resistor and the FET goes through 0.1uF to pin 3 of the LM386. Pins 2 and 4 of the 386 are grounded. Pins 2 and 3 are shunted by 0.01uF (located close to the chip). Pin 1 goes to the positive end of a 10uF capacitor, with the negative end going through 1K to pin 8. Pin 7 goes through 22uF to ground (negative at ground). Pin 6 goes through a 10-ohm decoupling resistor to the VCC line, and pin 6 is bypassed to ground through 220uF (negative at ground). Pin 5 goes through 0.1uF to pin 4. Pin 5 also goes through 100uF (positive toward pin 5) to one side of the speaker, while the other side of the speaker is grounded.

Pin 1 of the 555 is grounded. Pins 4 and 8 of the 555 are tied together and go to the VCC line. Between pins 1 and 8, located close to the chip, is 0.1uF. Pins 2 and 6 are tied together and go through 0.01uF to ground. Pins 2 and 6 also go through 100K to pin 7. Pin 7 also goes through 100K to the VCC line. Pin 3, the output, goes through 2.2K to the cathode of the LED (pin 2 of the H11F1). The anode (pin 1) goes to the VCC line.

Voltmeter "Multiplier" Circuit

Basically, this is a progressive voltage multiplier, with the chopper-comparator system always looking across the bottom resistor. The free section of the rotary switch--the one not used for on/off--is used to select the so-called "multiplier." Specifically, the free end of the FET channel goes through 11K to ground. The top end of this 11K (the connection to the FET) also goes to position 1 of the rotary switch. Between positions 1 and 2 is 100K; between positions 2 and 3 is 1meg; between positions 3 and 4 is 10megs. Subsequent positions are not used here; the builder has the option of adding to this network if desired.

The negative binding post (black) is grounded. The positive binding post (red) goes to the arm of this section of the rotary switch.

Milliammeter Shunt Circuit

The free section of the rotary switch--the one not used for on/off--is used to select the shunts. One end of each shunt is grounded, as is the negative binding post (black). The arm of the switch goes to both the free end of the channel and to the positive binding post (red). Across the binding posts is a protection diode (1N4003) with its cathode on the black one and its anode on the red one. The free ends of the shunts go to corresponding positions of the rotary switch as follows: 1K to position 1, 100 ohms to position 2, 10 ohms to position 3, 1 ohm to position 4, and 0.1 ohm to position 5.

Ruminating Over Layout

A common statement found in magazine articles and technical documents is, "Placement of parts is not critical, as long as good wiring practices are observed." What the deuce does that mean? Not only would an interpretation of this phrase be individual, but "good practice" varies with what you're building. For example, in high-frequency radio circuits, short direct leads--without mind to tracing at funny angles--are desirable. In digital circuits, orderliness is paramount. For circuits sensitive to noise, such as these meters, keeping inputs, outputs, and oscillators well isolated is of prime concern. Where there are digital items (like the 555 pulse generator) combined with high-gain audio circuits (like the LM386 amplifier) a "good practice" is to separate them by constructing them at opposite ends of the board.

Therefore, on our long skinny board, I recommend putting the amplifier near one end, and the offending oscillator near the other. The amplifier raises other concerns; experienced builders avoid having wires cross over from the output past the input (which sometimes leads to self-oscillation, or "feedback"). A cursory reading of the circuit informs us that the amp's input senses something elsewhere in the circuit. Putting the output side of the amplifier toward its end of the board would avoid this cross-over.

Remember that, at some later date, you might have to fix these things. Therefore, make a conscious decision to face all the chips the same way on the board. For example, let us place them with their pin 1 ends facing the VCC edge.

Picture the board with its "wiring side" facing up and with the VCC edge away from you; the pins of chips will poke up at you as they are inserted from the underside. The above decisions enforce the following arrangement: We have arbitrarily agreed that the "notch" between pins 1 and 8 of the LM386 will face away from you--toward VCC. From our previous work with this chip, we know that the input is on the right (pin 3, actually) and the output is on the left (pin 5). We have also decided that, in order to avoid a potential cross-over with the input lead, the output side should be near an end of the board--this would necessarily be the left end. We have agreed that the oscillator should be far away--this would logically be at the right end. What's left is to put the zener circuit and trim pot between the IC's; this smacks of "good practice," since that insensitive zener business constructs a natural barrier between the noisy oscillator and the sensitive amplifier.

The guy who wrote this article suggests that the H11F1 can share a 14-pin socket with one of the 8-pin IC's. Which one accommodates this? As we remember the amplifier, it's clear that it is surrounded by capacitors; it would be hard to visualize how that socket might be expanded for the new chip. On the other hand, review of the circuit shows that the 555 has nothing stretching between pins 4 and 5. Moreover, the oscillator is called upon to drive an LED in the H11F1 at a fairly high current; a long lead carrying the resultant electric field would lead to additional noise. Therefore, logic tells us to put the opto isolator immediately below the 555, sharing a 14-pin socket.

Finally, as soon as longer sockets are involved, this should set off an alarm bell that a wider board might be necessary. Following my own advice, my first model for the sake of this article was 1.9 inches (nineteen holes) in width. I built that one, only to find that room on the edges was wasted. I built a second with our old familiar board (broken off at the 16th hole), and it fit just fine.

The point of this section is: Yes, as you read a circuit with an eye to building it, you should try to visualize the parts, where their leads go in relation to other parts, etc. However, even with the limited experience of this series, you do not go into the fray unarmed. Many of the decisions you have to make are spelled out for you: by the size of the components, the number of pins which chips have, the disciplines of your conventions (which way you face chips and your favorite shape of board, for example), and the rules of "the Wiring Game" (see Part I, SKTF, Spring 1985). Also, don't let the imagined perfection of others fool you; you'd think after ten years of building close relatives to this very circuit, this author would have known precisely how wide his board had to be. He didn't.

Step-by-Step Construction

[Note: The electronics of these two instruments are very similar; therefore, in discussing the "on-board" wiring (and in discussing the cabinet layout), only one will be illustrated and discussed. If you want both instruments, "Play it again, Sam." (Of course, the input networks will be discussed separately.)]

The "Ruminations" of the previous section suggest initial decisions which are quickly restated here: Our familiar long-skinny board size will be used--measuring 1.6 by 4.5 inches, giving us 44 usable holes in the long direction and 15 in the short direction. Leaving a clear space on each end for screw holes, the amplifier will be built at one end, the oscillator at the other, and placement of the zener regulator and calibration circuitry--built in between these two main sections--can be dealt with later. A 14-pin socket will hold both the 555 and the H11F1 (or H11F3, whichever you have). It has been decided that the pin-1 ends of these sockets shall face VCC (as done in previous projects). We will mount the amplifier with its output side near the end of the board; as will be seen, this determines which end of the board is which.

Although rather unusual in building electronics with the point-to-point wiring technique, we know the size of our circuit board ahead of time. (In Part III, I discussed choosing a board which is too long, then breaking it to length after the project is finished.) The advantage in knowing the board size first is that mounting holes in the box can be drilled at the outset when this is easy to do.

Using the techniques discussed in part I, break off a board measuring 1.6 by 4.5 inches. Next, tape this board to the front panel of the box (one of the long sides) and drill clearance holes for 4-40 machine screws near each corner. I adhere to a "third hole rule"--the hole of the third column and third row from each corner; this gives screw holes in the perforated board a desirable physical strength. If you are meticulously consistent, these holes will match up no matter how you flip the board. If you are haphazard in positioning them, you will at this time have to declare which side, and which edge, of the board is which. (The component side is lying against the outside of the box, and the VCC edge is nearest the cover plate.)

Next, on the wiring side, install bus wires along each edge. For now, these can be anchored in corner holes with the idea that they can be cut shorter after connection points which will hold them in place are established.

Picture the board with its component side down, socket pins pointing up at you, and with each "pin 1" at the upper right of its socket. Bus bars should be visible on the side facing you. The one nearest you shall be "ground," and the far one shall be VCC. We have said that the amplifier should have its output at one end, and its input nearest the middle of the board where it will attach to other circuitry. In order for this to be so, the amp will be at the left end, and the oscillator will be at the right.

Building the Amplifier

Placement of the socket is governed by two constraints: First, we know from past experience that the output side (which will be facing the left end of the board as we look at the wiring side) takes up about an inch of room; we must account for this inch and at least another half an inch for the mounting holes. Thus, the socket should be placed a "generous inch-and-a-half" (1.6 inches) from the left end. Second, the vertical placement is always governed by the longest component going to a bus wire. In this case, a 22uF cap goes from pin 7 to ground. Well then, since my 22uF unit is a good 3/4 of an inch long, its leads, after being bent, would be about an inch apart; this would put its pin-7 end in row 10. Thus, the socket is placed with pins 8 and 1 in the 11th row up from ground.

The resultant socket placement is with pins 5, 6, 7 and 8 in column 16 (making pins 1 through 4 emerge in column 19), while pins 1 and 8 go into row 11 (making pins 4 and 5 go in row 8). The seasoned builder will put the socket slightly more than 1-1/2 inch from the end, with its pins 4 and 5 half way up from ground. Once this is accomplished, bend the pins outward and turn on the soldering iron.

If you plod along and follow the statements as they occur in the circuit description, the LM386 will be a troublesome chip. Its input connection (pin 3) is in the middle of power connections, grounding jumpers and bypass capacitors. I soon learned to take care of pin 3 first.

Fairly late in the circuit description, we find a 0.01uF bypass between pins 2 and 3 which is said to be "close to the chip." Install this capacitor right up against the socket (in column 20). Its two leads should be coming through holes 9 and 10 in this column, so that they line up exactly with pins 2 and 3. Stretch the pin 2 lead off somewhere to blank board and stuff it back through a random hole to anchor, or stabilize, this cap. Lay the pin 3 lead over its intended socket pin, cut this lead so that it just falls short of pin 6 across the way, solder it to pin 3, then cut it short (just beyond pin 3). Next, install the 0.1uF coupling capacitor a couple of holes away (in column 22, with leads in holes 7 and 9), anchor its free lead in like manner, make its run over pin 3 in like manner, and solder it thusly.

Once pin 3 has been taken care of, everything can go ahead in a less organized way. Disengage the pin-2 end of the 0.01uF cap and attach it to pin 2. Then, jumper pin 2 to ground with a piece of insulated nonstranded wire, running this jumper along the component side between the two capacitors. With a similar jumper, ground pin 4.

You will notice that the circuit says that pin 6 goes through 10 ohms to VCC, and that pin 6 is bypassed to ground by 220uF. Running pin 6 through a resistor to the VCC bus would be easy enough, but such a large capacitor would be hard to fit in the space between pin 6 and ground. There is a much easier trick which turns out to accomplish the same thing; in order to "license" this trick, however, you must notice that the demanding phrase "close to the chip" is not present. (In fact, where electrolytic capacitors are concerned, a constraint like "close to the chip" would be meaningless; the internal impedances of electrolytic capacitors are usually higher than the resistance of short lengths of wire, so no advantage would be gained by keeping wires meticulously short.)

[The trick is to give the LM386 its own VCC line, all "decoupled," filtered and everything. Somewhere about column 18 or so, we'll cut the VCC bus bar and insert the 10 ohm resistor. Then, the 220uF bypass can be installed nearest the end of the board where it is out of the way. This paragraph is in brackets because it would be inconvenient to do this now; we shall proceed with the understanding that we're going to do these things.]

Pin 6 can now be jumpered to VCC with a wire along column 15. Now, to change the subject, the leads of the pin-7 22uF cap will come through holes 1 and 10 in column 14 (two holes to the left of the socket. (The negative end is at ground, remember.) Soldering this caps ground end will stabilize it for a good connection to pin 7.

The bypass from pin 5 to pin 4 can now be installed just below the socket ("close to the chip," as they say). Crimp one end over the ground edge of the board to stabilize the cap. Arrange the other lead to comfortably reach its pin but not to cross over and bridge to the next pin up. Solder this first lead, then do likewise with the other; the solder connection on the first end will stabilize the unit.

If your output capacitor has "radial leads," you can position this just below the socket as well. Mine has "axial leads"; this means that I had to place it in a column to the left of the 22uF bypass. I installed it in column 11 with its negative end (which awaits a speaker wire) just short of the VCC bus (in hole No. 13). The positive end had to come up through hole 4 in the 11th column, whereupon this lead approaches pin 5 at an angle. Anchor the free end securely, since any rocking of a component with a long lead greatly aggravates the solderer.

Next install the series combination of 10uF and 1K between pins 1 and 8. I put my 10uF unit just above the socket and soldered its positive end to pin 1. Then, I put the resistor off to the side (with its leads in holes 11 and 14 of the 15th column). One resistor lead reaches straight for pin 8, while the other is arranged to cross the free capacitor lead somewhere above the socket.

Somewhere on the edge of the board above the amplifier, plug in the 10-ohm resistor. Mine is in the 15th row (along the VCC bus) with leads in the 18th and 23rd columns. Solder both ends to the bus bar. Then, cut out a quarter of an inch of the bus wire between these resistor leads. Finally, install the 220uF cap near the end of the board (in the 7th column)--with its negative end near ground--and solder its ends to the bus bars.

Testing the Amplifier

Insert an LM386 into the socket. Use clip leads to attach a speaker, one lead going to the free end of the 100uF output cap, and the other being grounded. Using clip leads, connect a battery to the bus bars (near the middle of the board, past the 10 ohm resistor). Touch the free end of the input capacitor (the 0.1uF unit off pin 3) and listen for crackles, humming, or squeals; these are signs of success. You can also use the test oscillator of Part III as a well-defined signal source; ground the cold test lead and run the hot one through 1 megohm to this input capacitor.

If nothing is heard, the continuity tests of "Trouble Shooting" in Part III will serve you well. Unplug the chip and try those.

Building the Chopper

The main disadvantage in using a common socket for two chips is the confusion that results in talking about them. The two rows of pins on the H11F1 are: 1, 2, 3, then 4, 5, 6. The 555 has 1 through 4 on one side, then 5 through 8 on the other. The 14-pin socket has 1 through 7 on one side, then 8 through 14 on the other. (Circuits made this way confuse sighted technicians as well; there's a guy working here who, when he sees that I'm sharing long sockets with more than one chip, leaves the room in a huff.) I propose the following conventions:

If I refer to a socket's pin, I'll state it as soc. pin 1, soc. pin 2, etc. About each chip, I'll give the chip number first--555 pin 3, H11F1 pin 4, etc.

Without a lot of figuring, we don't know just how much space the zener and calibration stuff will take in between chip sockets, so we will put the chopper socket as close to the end of the board as we dare. The output of the 555 faces this end; review of the circuit reveals that there are only a couple of components at this side.

A couple of components often means that two or three columns will be used (about 1/4 inch). Leaving 1/2 inch at the end for mounting holes, then 1/4 inch for components, we must conclude that the socket must be placed in the eighth column from this end (0.8 inches). As to its vertical placement, centered between the bus bars would be fine. On the other hand, boards look just a little neater if you can at least align one end of all the sockets in the same row--a minor point, to be sure. Nevertheless, this can easily be achieved here by placing the 14-pin socket with soc. pins 1 and 14 in the 11th row (just as LM386 pins 1 and 8 are in this 11th row(. Thus, soc. pins 1 through 7 come up through the 8th column from the right end (0.8 inches in), and pins 1 through 7 are in holes 11, 10, 9, 8, 7, 6 and 5, respectively.

Now, which chip goes in the top end of the socket? Review of the circuit brings to mind a bypass capacitor "close to the chip" between 555 pins 1 and 8. That settles that; 555 pins 1 and 8 must be at the end of a socket, which forces us to put this chip nearest VCC.

Although the power connections are mentioned first in the circuit, it is helpful to note that these are "end pins" (555 pins 1 and 8). Perhaps soldering these should be put off until interior pins are done. The H11F1 LED connections turn out to be "interior pins" now, and maybe these should be considered along the way? The following is my recommended order:

Soc. pin 4 goes to VCC via a jumper. (This is 555 pin 4, that the circuit states is tied to pin 8, which then goes to VCC; this builder has chosen to violate the circuit slightly and take this pin directly to VCC.) Soc. pin 5 (the H11F1 anode) also goes to VCC. Next, the 555 output and the H11F1 cathode are connected by a resistor. Soc. pin 3 goes through 2.2K to soc. pin 6; this resistor goes in the 6th column from the right end.

Next, the 555's pin 2 to 6 jumper can be installed; soc. pin 2 goes to soc. pin 12 via a jumper around the upper end of the socket. Make this jumper a little bit long so that it can eventually go around that pesky "close-to-the-chip" cap. Pin 2 then goes through 0.01uF to ground. This can be done in column 6 by putting the cap between 555 pins 2 and 1 (also soc. pins 2 and 1. However, if you prefer, one end of this cap can go to pin 2, with the other end taken down near ground and soldered there. This puts it on top of the 2.2K resistor, but that's a minor violation of the rules.

Pin 1 can now be grounded. Depending how you installed the 0.01uF capacitor, this jumper may have to stray toward the end of the board a bit, but this will assuredly be between the screw holes, so who cares?

Next comes a terrible "on-end" resistor between 555 pins 6 and 7. Anchoring this firmly before soldering is the secret of success. Select one of your nonprecision 100K resistors and bend one of its leads double where it emerges from the body; this will give you two parallel leads (one shorter than the other) which can fit into holes 1/10 inch apart. In the 13th column from the chopper end, and adjacent to soc. pins 12 and 13 (555 pins 6 and 7), insert this resistor in holes 9 and 10 up from ground. Bend the upper lead toward the amplifier, stuff it back through a random hole and crimp it firmly for stability. Solder the other end to soc. pin 12. Then, disengage the free end, bring it down on pin 13 in the usual way, and very carefully solder this to soc. pin 13. If your iron strays over and reheats the first connection, you can turn this resistor loose to flap in the breeze.

If you like soldering resistors on end like this, you can do so with another 100K between soc. pins 13 and 14. (The circuit says that 555 pin 7 goes through another 100K to VCC; but VCC is soon to be next door at pin 8.) If you prefer, you can lay this resistor down in the 14th column from the chopper end, making it connect between soc. pin 13 and the VCC line.

With a jumper in column 12 from the chopper end, connect soc. pin 14 (555 pin 8) to VCC. After this is done, the 0.1uF bypass cap can be placed at the top end of the socket and soldered between soc. pins 1 and 14 in the familiar way, taking care that its leads are not long enough to stray down to soc. pins 2 or 13.

Review of the circuit shows that H11F1 pins 4 and 6 (soc. pins 8 and 10) are all that remain. Soc. pin 8, which awaits an input wire, is an end pin, so we can leave this till later. Soc. pin 10 (H11F1 pin 6) goes to one end of a 47K resistor, as well as to the input of the amplifier. Where this resistor goes is of some concern; we don't want it to block the end pin when input connections are to be made. Therefore, I advocate lying it down in the chopper's 14th column--its upper end in the 7th hole adjacent to soc. pin 10, and its lower end down near ground, perhaps in hole 3 or 4. Solder its soc. pin 10 end; its other lead awaits a wire from the Braille pot.

A jumper must now go from the soc. pin 10 solder connection to the input of the amplifier. Make this a U-shaped jumper that can run along the ground edge of the board for the 1-inch-long lateral run it has to make. You might well ask, "Where can I plug this jumper in?' The pins either side of soc. pin 10 are not used. It's a slight breach of the rules, but the jumper can go either above or below the resistor lead, adjacent to an unused pin, then soldered against the 0.3-inch-long exposed resistor lead. The other end can go below the amplifier somewhere, where it can be arranged to cross the input capacitor's free lead.

Testing the Oscillator

The H11F1 is not wired to "chop" anything just yet, so this device cannot yet be tested. However, the oscillator is ready. Plug in a 555 at the upper portion of the 14-pin socket (nearest VCC). Connect a loudspeaker as before and power the board. Rest one finger on the 47K resistor (the one going to soc. pin 10 and which is jumpered to the amplifier. Feel the right half of the oscillator socket with another finger: you should hear a tone (which may moan and groan a bit as you change its frequency with your skin resistance. If you hear nothing, continuity tests are in order.

Remove the 555. Place a piece of 26-gauge solid wire in the negative test lead of your continuity tester; clip the positive test lead on the VCC bus. Probing the socket pin by pin, these should be your results:

[Note: the following list is of "soc. pins" only]

• Pin 14--should be a short (this is VCC for the 555).
• Pin 13--a low tone indicating 100K.
• Pin 12--a low buzz one octave lower than pin 13 indicating 200K (both 100K units in series).
• Pins 11 through 8--open circuits (the input capacitor of the LM386 goes to pin 10, but this is insignificant to the tester).
• Pin 1--a chirp indicating partial charging of large capacitors, settling to a fairly high pitch as the tester tries to supply power to the LM386.
• Pin 2--the low 200K buzz.
• Pin 3--open circuit (unless the H11F1 is plugged in).
• Pin 4--a short, VCC.
• Pin 5--a short, VCC.
• Pins 6 and 7--open circuit.

Building the Voltage Standard

The zener, whose anode is grounded, is otherwise in series with a resistor to VCC. Well, then, this could all be arranged in one column so as to span the board. The 470 ohm resistor is easy; its leads are bent at right-angles in the conventional way, whereupon it goes from the 15th hole (at VCC) down to the 11th. The diode needs a little more care.

My zener, like many small diodes, is made inside a glass envelope. You risk breaking it by bending its leads immediately at the points of their emergence from this glass capsule. Therefore, it is common practice to grasp each lead near the body with a pair of needle-nosed pliers, then bend the lead on the far side of the pliers (not bending on the "capsule side" of the pliers). The result is a somewhat longer component; the leads will be spaced farther apart than just the body of the diode. Thus, by the time you're finished, this diode might span 7 or 8 holes in that column.

The column I used is just beyond the amplifier circuitry--the 24th from that end and the 21st from the chopper end. The resistor goes from hole 11 to 15; it is soldered to VCC. (Make sure that the 10 ohm resistor in this line is not in the way; if it is, move the resistor down a hole.) The diode must be tested for polarity before insertion. When it is connected so that the tester gives the lower tone (lower than in the wrong direction), the end with the tester's negative lead goes in at ground, while the other lead (which was in the positive test lead) goes 7 or 8 up from there.

Arrange the cathode to cross the awaiting resistor lead and solder it; however, do not, at this time, trim the resistor lead, since you can use it later.

The screwdriver-adjustable calibration trim pot lies next to the diode assembly. It takes up two columns, since its arm is displaced from the two end terminals. With the screw pointing at the ground bus, I put the arm two columns away from the zener circuit, and I positioned the trim pot vertically so that the untrimmed 470 ohm resistor lead could be bent to reach this arm terminal. After bending the pot terminals outward to hold it in the board, solder the arm as described. (Of course, it is perfectly legitimate to connect the trim pot's arm to the zener junction via a small jumper.)

Strip, tin, and solder 6-inch lengths of stranded wire to the terminals of the Braille pot; in this circuit, all three terminals are used. The "bottom" of this pot (the terminal whose resistance to the middle one goes to zero when the arm is rotated fully counterclockwise) is grounded; find a convenient hole near the middle of the board and attach it there. The "top" of this pot (the opposite end of its winding) goes to either of the end terminals on the trim pot; this can be done along side the trimmer, or at its rear end (not in front of the screw, which would be inconvenient). The lead from the arm of the Braille pot should be plugged in somewhere that permits it to be attached to the free end of the 47K resistor off the chopper's soc. pin 10.

At this point, the chopper-comparator is complete. All it needs to work is a DC unknown voltage on the chopper's soc. pin 8--a voltage which can be covered by the range of the adjustable Braille-calibrated standard.

Building the Voltmeter's Selector Switch

Although a 6-position unit has been recommended because of its availability, only 5 positions are necessary (and one of these is the "off" position which has no connection). On the switches recommended, the terminals for the "arms" of the two poles are very hard to reach, so it is my practice to solder stranded leads to these terminals before doing anything else.

First, then, strip and tin two 8-inch stranded leads and solder them to the arm terminals. When soldering these, it is perfectly acceptable to reach between outer terminals with the soldering iron so as to land on these illusive inner ones.

It has been decided by the designer that positions 5 and 6 will become "off" positions; positions 1 through 4 become jumpered together and go to the red wire of the battery clip. To do this, hook a piece of solid bare wire through position 1, bring it around so as to encompass positions 2 and 3, then insert it into position 4 and secure it. Then, I recommend soldering 2 and 3 first. Soldering the middle positions first will make it easier to tell--by wiggling the ends of the wire--whether these are soldered securely; solder the ends afterward. (The biggest trouble with this procedure is making sure that the wire is in firm contact with the intermediate terminals. Forcing the wire to touch these lugs is essential before soldering.) The red battery lead--the one connected to the large snap--can be just wrapped around a segment of wire between jumpered positions, whereupon it is soldered.

Resistors are now to be installed between adjacent positions, which can be done in the order the circuit states. However, the resistor bodies are longer than the spaces between the lugs. What's to be done now? There are two solutions: One is to arrange the resistors in a larger circle--letting them stand out to the side on 1/2-inch "courtesy leads." The other idea is to double one lead of each resistor over where it emerges from the body (making the leads parallel, one shorter than the other), in which case their leads are very close together; then, they can be arranged any way you like. Whatever you do, make sure that these components do not protrude more than 1-1/4 inch behind the switch, or they will run into the metal bottom cover plate of the box.

Put the 100K precision resistor between positions 1 and 2; do not solder either lead just yet. Sharing position 1 is another 8-inch length of stranded wire; install this before soldering. Install the 1meg precision resistor between positions 2 and 3; solder position 2. Install the 10meg precision resistor between positions 3 and 4; solder both positions 3 and 4.

A review of the circuit reminds us that there is an 11K precision resistor, associated with this network, that goes to ground. However, there is no "ground" anywhere near this switch. (We could bring a ground over there, mounting a so-called "tie point," which is a mountable solder lug for this purpose, and grounding this point with a wire from the board.) However, this 11K appears across the chopper's input, so why not put it on the board. Thus, one more component is added to the voltmeter board; this 11K unit is put in a column just beyond the chopper socket (in the 13th column from the chopper end, its leads in the 1st and 5th holes). The bottom end is soldered to ground; then, the lead adjacent to soc. pin 8 (hole 5 in this column) is bent over against that pin and soldered.

The wire from position 1 of the switch comes through a hole just below the chopper socket; it is bent over to cross the 11K resistor's upper lead and soldered there. An 8-inch stranded ground lead is installed just below the chopper socket; this will go to the negative binding post. The positive binding post goes to the arm of the switch that selects the resistors. (We're a little ahead of our story, since the binding posts will have to be soldered after they are installed in the cabinet, as you remember.)

The black battery lead (the one connected to the small snap) goes to the ground bus on the board--anywhere will do. The arm of the switch associated with the on-off function goes to the VCC line--on the chopper side of the 10 ohm resistor.

Building the Milliammeter's Selector Switch

A point of concern to the builder of ammeters is that the shunt you are sensing must not include long wires. Thought the circuit may look as if you can do it, putting the chopper's input directly across the binding posts, adds enough additional resistance to the 0.1 ohm shunt to make the meter read higher than it should in the high-current position. Therefore, the chopper should go across the switch itself, with leads as long as you like then going from the switch over to the binding posts. This means that two sets of leads emanate from this pole of the switch; one set goes to the binding posts, while the other set goes to the board.

Thus on the pole devoted to the shunts, two 8-inch lengths of stranded wire are attached to this arm first. Then, on the on-off pole, one 8-inch length is attached to the latter arm.

Five of the six positions are used here. After attaching 8-inch leads to the arms, positions 1 through 5 of one pole are jumpered together; a solid bare wire is inserted into positions 1 and 5 and is made to encircle positions 2, 3, and 4. Position 3 is soldered first, then tested. Then positions 2 and 4 are soldered and tested. Finally, the end positions are soldered. The red battery lead is wrapped around one of the wire segments between lugs and is soldered.

In this case, one end of each resistor goes to its assigned position, while the far ends are joined together (very reminiscent of the switch arrangement in the resistance bridge of Part IV). Solder the longest resistor first (this will probably be the 0.1 ohm unit on position 5). Bend the free end of this first resistor at right angles and put a gentle curve in it so that it follows the same curve as the switch terminals. All other resistors go between their assigned switch positions and the curved lead off the first resistor (one end of each being crimped in the appropriate switch lug, and the other being wrapped around this curved wire). This assembly of resistors cannot extend more than 1-1/4 inches behind the switch, or they will hit the metal bottom plate of the box.

As stated in the circuit: position 5 goes to 0.1 ohm, position 4 goes to 1 ohm, position 3 goes to 10 ohms, position 2 goes to 100 ohms, and position 1 goes to 1K. One of the wires from the arm of this pole goes to soc. Pin 8 on the board (the free end of the H11F1's channel). A stranded wire from the junction of the resistors goes to ground on the board. Eventually, the arm of this pole goes to the positive binding post, and the junction of the resistor ends goes to the negative binding post. (The binding posts cannot be soldered until they are installed in the box, remember.)

From the junction of resistors, the ground wire to the board can be plugged in anywhere--just below the chopper socket is a good place, for example.

Other than the on-off function, the board need not be directly involved with the selector switch; two 8-inch input wires from the board will go directly to the binding posts. The ground wire can be plugged in anywhere, just below the chopper socket is a good place (the other end of this wire eventually going to the negative binding post).

The hot input wire goes to soc, pin 8 of the chopper; this can come through a hole in the 13th column from the chopper end and adjacent to this pin. Before soldering this latter lead, care must be taken to anchor it firmly--with an alligator clip or some such device--so that it won't be easily bumped over to a neighboring pin. (This hot lead eventually goes to the positive binding post.)

For the sake of completeness, it should be stated here that the protection diode is soldered directly to the binding posts, not necessarily to the switch or on the board (although those other alternatives would work okay). When properly installed, this diode should make your continuity tester beep when it is clipped across the binding posts (positive to positive and negative to negative). (This test should be performed with no shunt in place; i.e., do it with the switch in the "off" position.

The black battery wire goes to ground on the board (anywhere). The arm of the on-off switch section goes to the VCC line--on the chopper side of the 10 ohm resistor.

Preparing the Cabinet

The cabinets for both instruments are identical. The metal cover plate will be considered the bottom of the box. The short sides will be the left and right ends; the long side toward you will be considered the "front" of the cabinet. As per the procedures and techniques described in Part IV, the following placement of parts is recommended:

The main pot is mounted on the top in a 3/8-inch hole 2-3/4 inches from the right end and vertically centered (2-1/8 inches from the front). Also on the top panel, the rotary switch is mounted 1-1/2 inches from the left end and is vertically centered (2-1/8 inches from the front). For both of these controls, provide holes for their respective locating tabs; this will prevent them from twisting during operation (see Part IV for details).

The speaker is located on the right panel. The board is mounted ot the front panel with its calibration screw facing the metal cover plate; this is done using 3/4-inch 4-40 machine screws and nuts, with 1/4-inch-long spacers holding the board away from the cabinet. The 9-volt battery is mounted on the cover plate (see Bob Trottman's fine article on the subject in this very issue).

The binding posts are mounted on the rear; they are spaced at 3/4 of an inch between centers so that they can accept standard two-prong banana plug fixtures. In the voltmeter, a ground wire from the board goes to the black one, and a wire from an arm of the selector switch goes to the red one. In the milliammeter, a wire from the junction of all the shunts goes to the black binding post, and the extra wire from the switch arm goes to the red one. Finally in the milliammeter, a protection diode is shunted across the binding posts--in such polarity that the diode will conduct when the instrument is turned off.

Installing the Braille Dial and Pointer Knob

Braille dials that match the main pot have been provided. When Thermoformed, a washer was placed in the center of each dial; this can be converted into a "locating dimple" to help with positioning of the dial on the top panel of the box.

Place the dial with its center approximately over the 3/8-inch hole intended for mounting the pot. Press down on the "washer" portion until it snaps downward to fill this hole. (In other words, this center raised portion can be inverted to protrude in a downward direction instead, whereupon it will fill the hole and help "locate" the dial.) Next wipe the underside of the dial clean with isopropyl alcohol; do the same to the top of the box.

Apply strips of double-sided cellophane tape to either the dial or to the top of the box. Using the inverted dimple to first locate the hole, lower the edges of the dial into place. Then, with a small knife blade, carefully "pare" out the dial plastic from around the mounting hole (the plastic of the box is very soft, so be careful).

The pot comes with an "inside star" washer which is of little use here. It is best if we had a smooth washer to go on top of the dial, under the nut. As it happens, you might find such a smooth washer as standard equipment with the rotary switch; steal it for this other worthy purpose.

When installing the pointer knob over the Braille scale, it is important to remember that the mechanical rotation of a pot may slightly exceed the so-called "effective electrical rotation." In other words, a couple of degrees at each end constitutes a "dead portion" over which no change of resistance occurs. It is necessary to find the position of "true electrical zero" on the pot, then install the pointer knob so that this coincides with the mark for zero on the dial.

Turn the instrument on with the main pot fully counterclockwise. Slowly advance the pot in the clockwise direction until the point is found where the sound just starts to increase. Now install the knob so that it points at zero when the increase is just about to happen.

Calibration

Calibrating the Voltmeter

Batteries are not bad voltage standards when they are fresh. A new zinc-carbon cell has an open-circuit voltage of 1.55 volts (this would "null" just past the second minor division on the 10-volt scale). Freshly charged nicads are 1.25 volts per cell; a group of four in series would give you an even 5 volts (mid scale on the 10-volt position, position 3). You can even use its own zener as a standard; touching the cathode of the zener with a clip lead from the positive binding post will present the meter with a nominal 6.2V.

Whatever your standard, set the voltmeter to the smallest range that would accommodate this measurement (the 10-volt range for all the above examples). Set the pointer knob to the reading you expect (slightly beyond the sixth major mark for 6.2V, for example). Then, connect the standard and turn the calibration screw until a "null" of the tone is obtained.

Calibrating the Milliammeter

Precise current sources are not as easy to come by. An obvious method would be to compare its reading with a known meter. However, a trick is up this author's sleeve.

With no shunt applied--when the switch is in the "off" position--this instrument is a 100 millivolt meter of practically infinite impedance. (The protection diode does not come into play until the voltage on the terminals exceeds about 600 millivolts.) Inside the companion voltmeter is an identical 100 millivolt standard which we calibrated in the previous section. Therefore, open up the voltmeter and measure the voltage across its Braille pot with the milliammeter as follows:

First, the milliammeter must be powered while in the off position. Use a clip lead to connect the jumpered switch lugs to VCC. (VCC is not terribly convenient to find with the board screwed down, but it can be found either at the arm of this same switch section, or at the chopper end of the 10 ohm resistor in the VCC line.) Next, connect the negative milliammeter binding post to the bottom of the voltmeter pot, and the positive binding post to the top of that pot.

Turn on the voltmeter; leave the milliammeter's switch in position 6. Turn the milliammeter's Braille pot all the way clockwise. Then, adjust the trim pot with a screwdriver until a "null" is obtained.

Embellishments and Improvements

On the surface--as you might say--it seems that one problem with all these instruments is that the pointer knobs are never close enough to the Braille scales. The remedy for this is to construct a plastic spacer under the top panel made from a 2-inch square piece of scrap (salvaged from a similar box which didn't survive the war, perhaps). Drill a 3/8-inch hole in this new piece and cement it to the underside with Pliobond cement. (Use the pot's bushing or a 3/8-inch drill bit to force the holes to coincide. Then, redrill the hole for the "locating lug.")

You may find that these instruments aren't loud enough for your liking. If not, change the resistor on pin 8 of the LM386 to perhaps 470 ohms or less.

As suggested, both instruments could be built into one box. Obviously, the on/off switch will be an extra toggle. The two poles of the rotary switch can accommodate both input networks; however, switching of the chopper-comparator inputs between one and the other is not entirely trivial.

First of all, note that I suggested that the 11K resistor in the voltmeter network be placed on the board. This can no longer be, since this resistance must disappear when going into the ammeter mode. Therefore, this resistor must go from the arm of its associated switch to ground. The hot input of the chopper-comparator system must be switched between the arm of the voltmeter switch and the arm of its ammeter companion.

Separate terminals should probably be used for the two meter functions. This is true because a protection diode limits the voltage across the ammeter. (Separate terminals tend to keep the absent-minded technician from throwing the switch in the wrong direction and putting an ammeter across an unforgiving voltage supply.)

By itself, the chopper-comparator circuit comprises a voltmeter with practically infinite input impedance (when it is adjusted for a null). The 11K input impedance seen on the 100mV position could be completely dispensed with if appropriate switching were provided. Beyond this, the 1V position could be made to replicate this sophistication if the voltage on the Braille pot were stepped up to 1 volt in the 1V position; however, the instrument would be so loud that you would want to disconnect the resistor from pin 8 of the 386 for this mode. (Boy, now we're really getting into some fancy switching requirements.)

The multiplier network of the voltmeter is not exactly right; it is off by about 1%. This could be cured by putting a 110 ohm resistor in series with the 11K unit. (The bottom resistor should actually be one-ninth of its series neighbor.)

Right now, the milliammeter suffers from non precision resistors. If you find a supplier who is willing to sell 1% units to ordinary clods like us, please let us know.

Measurement of AC milliamps is not easily accomplished. Diodes, by themselves, do not work at voltages around 100mV. However, if you look back through Al Alden's articles on operational amplifiers (specifically, Part II in SKTF, Spring 1983), you will find circuits of so-called "ideal rectifiers" which could be used. (It would probably be wise to convert the chopper-comparator circuit to measure 0-to-1 volt to avoid offset problems and so forth; the RC network between pins 1 and 8 of the LM386 would then be eliminated to cut the volume down.)

If you are willing to stick to voltages over 1 volt, AC voltage measurements are more easily accomplished. (This would only work for the top two ranges--10V and 100V.) The catch is that you will be measuring "peak" values; therefore, multiply each reading by 0.707 to get the RMS value of sine waves. A special probe needs to be built as follows:

A double-banana fixture is used (Pomona No. MDP-0). Across the prongs of the plug is mounted the parallel combination of 0.47uF and 1 megohm (a 5% 1/2-watt resistor is used, and a Mylar or "metalized" capacitor whose rating is 200 volts is employed). The negative test probe goes to the negative side of this plug (usually marked with a bump on one side). The positive test lead goes to the anode of a diode (1N4003), with the cathode going to the positive banana prong.

It is worthy of mention that the chopper-comparator is ideally suited to reading visual meters. Thus, if you find a visual "analog" (not digital) instrument that has capabilities you want, use this circuit to convert it for your use (in rare cases, you may have to put additional resistance in series with the trim pot so that lower comparison voltages can be accommodated). (For more details, see SKTF, Spring 1981.)

Addresses

• Pomona Electronics
  1500 East Ninth Street, Box 2767
  Pomona, CA 91769
  (714) 623-3463
• Jameco Electronics
  
  1355 Shoreway Rd.
  
  Belmont, CA 94002
  (415) 592-8097
• Mouser Electronics
  11433 Woodside Avenue
  Lakeside, CA 92040
  (619) 449-2222
• Digi-Key Corp.
  P.O. Box 677
  Thief River Falls, MN 56701
  (800) 344-4539

FINALLY, A 9-VOLT BATTERY HOLDER

This fine idea was submitted by Bob Trottman of London, England. The following three paragraphs comprise his letter:

"You can buy lengths of enclosing channeling in which one can hide and protect household electrical wiring. This is called "trunking." It is rectangular in cross section; one side can be removed so that cables can be placed inside the trunking and the removable side of the rectangle can be snapped back into position. The bottom section has lips which match molded ridges in the lid.

"In the United Kingdom, the trunking is made of plastic, and it comes in 6- or 12-foot lengths. Trunking comes in different cross sections. I find that the one that measures about 1/2 inch deep and 1 inch across the sides makes ideal battery holders.

"When I want to mount a 9-volt transistor radio battery, I cut a 2inch length of trunking, remove the lid from this length, and fix it where I want it. The battery will fit tightly between the jaws of the open side of the little length of trunking. The battery is held in place in a nice firm grip, and you can fix this "holder" any way you please."

By George, I've been waiting for that for 20 years! Good for you, Mr. Trottman. The resultant assembly is a sleek-looking battery compartment whose cover can either be slid on from one end, or snapped down, with a fair amount of muscle, from the top.

The "trunking" referred to can be gotten in electrical supply stores -- unfortunately, only stores of major size, since this big stuff is made for large wiring installations. The best I found here is so-called "twopiece surface-mount metal raceway" by the Wiremold company. Unfortunately, I haven't found a plastic version of appropriate size; what I'm currently using is fairly heavy steel (0.025 inches thick). You have to buy the two pieces separately: the Wiremold 2000B is the bottom half, while the 2000C is the cover. Together, they cost $4 for a 5-foot length.

Outside dimensions are listed: 1-9/32 by 3/4 inches. The battery fits sort of loosely in the assembled pieces. This means that some way of keeping the battery from sliding out the ends is in order. This can be accomplished by using long mounting screws at either end (5/8 or 3/4 inch screws). My favorite assembly length is 2-3/4 inches, giving me room to mount the bottom section by its ends. There are prepunched holes-"knockouts"-- every 1-7/8 inches in the bottom section; these should be avoided when drilling smaller holes for mounting.

Cutting the material is a bit of a trick, especially since the individual pieces are rather awkward to hold in the vise. The best idea seems to be to snap the two halves together first (starting at one end and mashing them together); then saw them both simultaneously. (Your little pieces match, that way.)

I list two addresses for Wiremold here. Buying directly from them is unlikely, especially with their required $150 minimum order. Nevertheless, they will no doubt be able to send you to local dealers.

• The Wiremold Co., Electrical Division, 60 Woodlawn St., West Hartford, CT 06110-0639; Phone: (800) 621-0049, or (203) 233-6251.
• Wiremold Canada, Inc., 5915 Atlantic Dr., Unit 5, Mississauga, Ontario L4W 1S4 Canada; Phone: (416) 677-393