The Smith-Kettlewell Technical File

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

William Gerrey, Editor

Issue: sktf-Winter-1990

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

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 Details regarding products, suppliers, and other contact information are original and may be outdated.

Questions about this archive can be sent to


Notes on Tactile Drawings: Back to the Drawing Board

The Smith-Kettlewell Battery Tester: Where Silence Is Golden

Bill's Box of Knowns


by R.J. Trottman, G3 ILU


In the United Kingdom, the Royal National Institute for the Blind (RNIB) is the principal publisher of Braille and the only producer of embossed drawings. I have seen examples of such dating from well before World War I. Most of these drawings were of maps, anatomical diagrams, and simplified sketches, as well as electrical and electronic circuits. Therefore, virtually all the techniques described here have been used by or published by RNIB.

In view of the success of the plain language descriptions of circuits of all kinds in SKTF, and formerly in the Braille Technical Press, it might be asked, "Who needs embossed or tactile schematic diagrams anyway?" There may arise a situation when being able to scribble a quick sketch of a circuit design for a sighted colleague or friend could be very useful. I find that many sighted people have difficulty grasping a verbal description of a novel circuit. If a blind technician can learn to follow a tactile drawing, a sighted person could draw a diagram quite simply.

In my case, I would answer that I received my basic education in electronics very largely through the medium of such diagrams. I believe that diagrams greatly help the study of circuit function and analysis and in fault finding. Furthermore, there are other drawings of value in the study of science: charts and curves, polar diagrams and vector diagrams that should be reproduced in a tactile form. Therefore, read on!

[Editor's Note: I would never quibble with the logic of the above. We need the commonality of language that enables us to communicate with sighted people. As for mathematical graphs and vector diagrams, there is no substitute for getting our hands on, at the very least, representative samples of these.

Creating tactile drawings for SKTF would send me into a loop of assessing proof copies and such. What I may do some time is Thermoform schematic symbols so that those who deal with readers can better communicate with them. For presentation of full schematic diagrams, I am in agreement with Bob Gunderson--they're not so efficient. When mathematical issues come up, I'll just have to "wing it" somehow, probably sending separate sheets.]

Between 1926 and 1972, RNIB published five stereotyped books on radio containing reproductions of circuit diagrams and other illustrations (stereotype, here, means machine-printed for sale). The graphics were embossed, by hand, on zinc plates like those used for normal Braille printing. Since World War II, RNIB has added a few textbooks on electronics for their Students' Library. These were transcribed by volunteers in the inter-line format, and the figures were drawn manually by the RNIB in-house draftsman.

It used to be RNIB policy to copy pretty exactly the diagrams in the radio textbook chosen for transcription, but since World War II, a Braille-based convention has been used to represent certain components. Capacitors are shown in the usual way: two short parallel strokes with connections coming off the center of each parallel. Resistors are depicted by a series of at least three Braille letter Es or Is, which suggest the angles of the old zig-zag resistor symbol. Inductors are represented by an alternating series of close-spaced Braille Cs and Gs, again suggesting the loops of a coil. I suppose that a special guide is fabricated to apply those symbols to the diagram, and the same device would be used to write component identifications so the symbols could be placed in the lower half of the cells. I guess that most other symbols may be drawn in conventional form, including circles for tube envelopes and semiconductor cases, etc.

I think that component symbols are drawn or "written" first in their proper place on the paper, and then connected by straight lines. If the lines have to take a circuitous route, then changes of direction are made at right angles. Where lines cross without indicating an electrical connection, a small arc or kink is executed to suggest that one wire jumps over the other. It must be mentioned that Braille cannot be written on the reverse side of the page containing a tactile drawing.

At the beginning of each volume of a transcription, it is usual to show a collection of samples of symbols to be found in that volume, especially any introduced for the first time.

I think that the first two factors to be borne in mind about tactile diagrams are that they need to be much larger than they would normally be, and some of the finer detail may have to be sacrificed. It must be remembered that the sense of touch is not as acute as sight, and room must be found for Braille identifiers which may take up more space than the symbol they mark. I would estimate that a tactile drawing of the classic Wheatstone bridge would occupy about two-thirds of a Braille page. It might be possible to squeeze a three-cascaded transistor amplifier on a full page if there is not too much in the way of feedback circuitry to be shown.

Another factor to be considered is the assumed level of astuteness of the user; a beginner may need a larger scale drawing, and an experienced person may tolerate a smaller, more closely spaced and detailed execution.

Generally, any illustrative diagram is likely to require proportionately more space than it does in printed text. Therefore, a tactile diagram may occupy more than a single page of a particular Braille publisher's preferred page size. Drawings may be divided between facing pages or sectionalized into consecutive pages, but Murphy's Law will ensure that important features will be crammed at the page edges or in the cleavage between pages! When the splitting up of a drawing is necessary, it must be made as easy as possible for the reader to follow a particular line from page to page. I have seen diagrams drawn on extra-long sheets of paper which were normally folded into the book, but damage soon developed in the bite of the fold and the crease weakened the paper.

Old Traditions

The oldest and most long-standing method of producing embossed drawings involves the use of a tool holding a serrated wheel--a "spur wheel." During a visit to RNIB, I was shown a selection of these. They consist of a round metal handle about the size of a ballpoint pen, with a small wheel having a serrated rim mounted on a transverse bearing at one end. The wheels fitted on two of these tools were about 1/4 inch in diameter, while another bore a wheel of at least twice that diameter (that is recommended for drawing curves and circles). These are suitable to be fitted to draftsman's compasses which the RNIB also offers. The Institute sells a set of seven different wheel heads that fit a common handle which is recommended for drawing maps and similar complex figures, so they should be very useful for drawing schematic diagrams as well as associated material.

To make an embossed drawing, one lays a sheet of manila (Braille) paper upon a so-called geometry mat, which is a sheet of resilient rubber about 1/8 inch thick and large enough to accept the largest size of standard paper used for books in the UK (25.5 by 34 cms). Using one of the spur wheel tools just described, one draws a line by allowing the wheel to roll on the paper which, if done with just the right pressure, will produce a row of tiny, closely adjacent dots on the underside of the paper. This means, of course, that the drawing must be executed in "mirror" fashion.

German Film

The RNIB imports from Germany a special plastic foil which they sell as "German Film." It comes in sheets somewhat smaller than the largest manila Braille size, but still adequate for many drawings. A sheet of this film is laid on the geometry mat and the drawing is made with a scriber in the form of a dummy ballpoint pen complete with a pocket clip. The great thing about this film is that the drawing appears on the upper surface, so one can monitor one's progress and not have the problem of mirror imaging to think about.

The ballpoint pen creates a crease in the thin plastic film. Because the plastic is thin, this crease can be felt from either side, as long as it is resting on a flat surface. The distortion, or crease, most often ends up lying on its side, but the illusion that this crease is a raised line is a good one.

Immediately as I began drawing on the film, I was struck by the clarity of the line and the smoothness of the pen. In what seemed no time at all, I did a freehand drawing of a very basic crystal receiver, complete with antenna and ground symbols, inductor, a solid-state diode, a couple of capacitors and headphones. I had done nothing like it for 25 years, and I was quite excited by the result. I do not think that one needs a special pen--I could make good lines with a Braille style. The film is strong but very flimsy, so if one wanted to keep any work, it would be necessary to mount it on paper. Anyone not living in the UK may have to obtain the film direct from Germany. [Similar equipment is available from APH as the "Raised-Line Drawing Kit."]

A blind friend of mine, who worked for a large British radio manufacturer for twenty years, told me that he drew large diagrams using a rubber-faced board fitted with a straight-edge, together with a commercial brand of plastic photographic sheet; he and his sighted workmates could draw diagrams on a surface much greater than normally available with anything from RNIB. Also, the membrane accepted ink from an ordinary ballpoint pen. These drawings had a much more pleasant feel than other plastic films. The material was strong and as stout as normal paper, and it did not suffer from the build-up of static electricity (which is the bane of many plastic films). Unfortunately, my friend changed careers some years ago, so he does not remember the name of the supplier. I suggest that interested readers make inquiries locally.


This method of producing electronic diagrams was devised by the late G. F. Day (ex-Royal Navy blinded veteran). The term "Brinkprint" was derived from Braille and inkprint because of the elements of Braille and print practice employed in the system. The "drawing" is done with a Perkins Brailler fitted with a second ratchet and pawl on the roller so that the paper can be moved one-half a line space.

To produce a horizontal line, one simply writes a row of dots 1-4, 2-5, or 3-6. To produce perpendicular lines, one writes a series of dots 1-2-3 or 4-5-6 in line, one above the other. Using the extra ratchet and pawl--which works in conjunction with the original ratchet and pawl--one can get an unbroken line. Lines tend to be at right angles, although it is possible to produce lines at about 45 degrees to the horizontal; these are not quite as satisfactory.

Creating perpendicular lines is a bit tedious as you have to backspace as you move up and down. It is possible to get the same result with an unmodified Perkins by simply balancing the roller halfway between lines, but there is the risk of the roller slipping. There is a slight difference between horizontal and perpendicular lines, but it can be ignored.

It is quite easy to create rectangles which can enclose Braille signs; therefore, the simplest use for the system is to make block diagrams.

I will leave it to the reader's imagination or ingenuity to develop the above. With Day's method, it is practicable to produce good schematic diagrams. Good approximations to all the major component symbols can be created with the Perkins Brailler.

When making an electronic schematic, one works from the input at the top of the page downwards; then, to read the diagram, one turns the page counterclockwise through 90 degrees to bring it into the conventional orientation. This is quite convenient as most components such as resistors, inductors, and capacitors appear in the vertical plane and are, therefore, easier to "write" horizontally in Brinkprint. The symbols for inductors and resistors are the same as those described earlier.

Home-Grown Tactile Illustrations

There is an alternative method of creating a tactile drawing.

A thin but rigid base is chosen of the same dimensions as the intended copy. Over the whole area of the base a mild adhesive is applied. The adhesive should not be a glue, but should have the "stickiness" of that found on insulation tape. In fact, if you can find double-sided adhesive tape which is wide enough to span the base, it will probably be ideal. On a prepared base, a diagram may be stuck on the sticky surface. Most lines and symbols could be fashioned with annealed wire. So that they stand out, the symbols should be made from a somewhat heavier gauge of wire than that used for connecting lines. Small beads could be used to show a dotted line and even to form Braille letters, although it is usual to write Braille on pieces of paper just large enough to accommodate the text, then apply these labels as close as possible to the symbol. The creation of a "drawing" can challenge the ingenuity of the maker, especially in the case of illustrations.

When the illustration is completed, the exposed adhesive must be neutralized by applying a powder like French Chalk or talc.

The foregoing description of making illustrations is OK when a freehand drawing is attempted, but if it is desired to make an accurate copy of an actual diagram, one can use a transparent base covered with a translucent adhesive. A plate of clear glass is mounted over a good light source, and on the glass is laid, possibly, a photographically enlarged copy of the diagram. Upon this is placed the base. If the light is bright enough, an image of the drawing should be visible through the adhesive. Then, whatever material you choose to make the tactile illustration of, it can be laid down according to the illuminated drawing.

I have been told of someone who used boiled spaghetti to form lines! I guess they had the leftovers for lunch!


I suppose that most Braille users are familiar with products of the Thermoform process of duplication, even if they have not used or seen the hardware involved. Briefly, a sheet of Braillon is placed on top of a "master" and is grilled under a fierce heat and some pressure for a few seconds, during which the Braillon permanently takes on the contours of the "master." Even if you do not like the feel of Thermoformed Braille, it must be admitted that copies are of good definition. Leaving aside the capital cost of Thermoform equipment, it is probably the cheapest method of copying embossed material. Spur-wheel drawings on manila paper can be successfully copied by Thermoforming. Naturally, if another method of creating an illustration is used, the master must be made of materials that will stand up to the heat of the Thermoform apparatus. German Film drawings and previously made Thermoform copies cannot be used as masters.

The "Minolta" System

Minolta is a Japanese photographic company name, but it has now come to stand for the latest method of directly reproducing tactile drawings. Being famous for years in the manufacture of professional-grade cameras, the Minolta company is now known for its photocopying equipment as well.

The tactile Minolta system depends on a special paper which is impregnated with millions of minute "capsules" that swell when subjected to a powerful light source. The largest size of this paper that I have seen is 35 by 26 cm.

The master drawing is executed in black on a sheet of ordinary paper. The master is placed on the "capsule" paper and both are inserted into the Minolta copier. The light inside the copier is reflected by the white background but is absorbed by the black lines of the drawing; this activates the capsules immediately underneath, causing them to dilate and form a ridge in correspondence; it's a sort of controlled blistering. There is no sign on the reverse side of the copy of anything on the upper surface.

The original drawing will probably have to be done by a sighted person, and it is likely that the drawing will have to be enlarged and simplified with non-essential details being left out. (This is true for any tactile copy.) The running-off of copies is very quick and easy, but the apparatus and the capsule paper are expensive.

The resulting copies are very good. To me, the drawings have a softer feel than Thermoform copies, but the detail is as clear. The tactile quality is more pleasant and "faster" than Thermoform.

Hi-Tech Graphics

For those at the cutting edge of microprocessor technology, the foregoing will doubtless seem very passe. I have seen embossed graphics of various waveforms produced by computers and Braille printers. Mechanically, these printers embody refinement of the Brinkprint system. Many character printers have a print head which consists of only three vertical dots. The paper is rolled up or down by a stepper motor which, as you can imagine, can place the paper anywhere in the vertical dimension. Then, too, some of these printers do not punch into dimples in a die bar; the "die" consists of continuous horizontal grooves which can accommodate the printhead styli anywhere in the horizontal dimension.

With such printers, it is possible to place dots anywhere on the page. As a further refinement, firmware in some of them makes it possible to rotate the printing process by 90 degrees. Although printing takes longer in this mode, they can write Braille vertically over the length of two Braille pages, thus allowing for an 80-character line, or schematic diagrams that are two or more pages wide can be made.

With all of this, it is not an automatic process to translate visual computer "graphics" into Braille. Thus far, special software written for the expressed purpose of making raised-line illustrations must be used.

It must ultimately be possible to write a computer program to enable one to type-in a verbal description of a circuit, strike the print key, and lo -- the embosser roars, and out comes a sheet of manila paper, or whatever, bearing a perfect tactile schematic. OK, all we need is the hardware, the software, the room and the money!

The chances of accomplishing this would be greatly increased if we wrote circuit information in a standardized notation. A shorthand format for describing schematic diagrams, such as the "Braille Linear Representation" to be described, might be the best tool to start with.

Braille Linear Representation

By devising a shorthand notation, it is possible to put circuits into Braille and save the tedium of wading through all that verbiage! In fact, a well-thought-out code does exist.

In 1962, RNIB published a Braille-based code for representing electrical and electronic circuit diagrams as the fourth, and final, part of their "Braille Science Notation." Ten years later, an improved and expanded version was produced, and a new edition is promised for Summer, 1990. The code is a development of that in a pamphlet describing the "SladkyBuczkowski-Benham" code for brailling circuit diagrams which was brought out in the USA at an earlier date.

The 1972 edition of the code is comprehensive and straightforward. There are fourteen "rules" which are quite easy to master.

Each component is represented by one or more letters, just as you would identify it on a diagram: C for capacitor, L for inductor, R for resistor, etc. Serial numbers follow the component letter in the lower half of the Braille cell, and "V" immediately after a component indicates that it is variable. There are three one-cell signs to indicate parts of multisection or ganged components, such as variable capacitors and banked switches. State, positive or negative; hot or cold; pin number--these are marked with standard indicators. Dot 5 close to a component letter shows that there is a connection to and through it. Parallel connections are clearly indicated, and so are mutual magnetic couplings. There are conventions for writing-in component values and type numbers.

That is a very brief description of a most practicable method of reducing long-winded, plain language expositions of circuits to a manageable compactness. One could, perhaps, devise one's own system, but with an established standardized code, it would be practicable to exchange circuits with other visually handicapped technicians.

The Bottom Line

The subject of tactile graphics is controversial among some blind people and their associates. Some congenitally blind individuals find no value in embossed drawings; yet others, especially adventitiously blind folks, receive much help and pleasure from them. I think that the latter opinion is supported by experience in the field of Fine Art and the blind. Also, it must be remembered that most "blind" persons do have enough sight to see a tactile illustration perforce. I hope that the foregoing survey will stimulate interest among the "grass roots," and a more authoritative treatise on the topic of tactile technical drawings may be forthcoming.

Address List

American Printing House, 1839 Frankfort Ave., Louisville, KY 40206: Phone: (502) 895-2405.

Royal National Institute for the Blind (RNIB), Overseas Inquiries, P.O. Box 173, Peterborough, Cambs. PE2 0WS, England.



The simple output from this instrument allows the user to make judgments about voltages in the very narrow range from 1V to 1.5V. The duty cycle of a recurring tone gets longer as the cell gets weaker until, at the lower limit, the beeps run together to produce a continuous tone. Good flashlight cells and good 9V batteries will cause the tester to emit short "pips," or no beeps at all if they are very fresh.

Introductory Editorializing

"Battery testing," as most people would like to perceive it, turns out to be a hopelessly undefined task. Different kinds--"alkaline," "zinc-carbon," "mercury," nickel-cadmium--have vastly different discharge characteristics and different terminal voltages. Some examples are: In top shape, a nickel-cadmium cell will never test "as good" as other types, even though, in terms of energy storage, it can run something just as well. A mercury cell will check out to be just fine until moments before its death. A so-called "heavy-duty" zinc-carbon cell may appear dubious, but by way of absorbing its hydrogen bubbles, it may significantly revive overnight.

For all but the mercury cell, one could build a monstrosity which would plot the voltage as a function of varying load. Knowing what type you are testing, this curve could then be compared with those in battery books, and the remaining energy would then be somewhat predictable. Then, in order to predict battery life, you would have to know something about the current drain of the appliance you are powering (how loud you play the radio at the beach would make a difference).

There are three practical alternatives to predict how much life is left. (1) Keep a log on the use of your battery appliances (no thanks). (2) Retrieve the motor from a broken toy, see how fast it runs on a fresh cell, and compare this with how it runs on the cell in question. (3) Use the S-K battery tester.

In the class of testers that generalize the battery checking problem, the Smith-Kettlewell battery tester is as good as they come.


Two versions of this instrument are possible. The one we built uses the Radio Shack "Handy Checker," No. 22-096, which has an adjustable cell holder and applies load resistors to single cells and 9-volt batteries. The other version, made without the Radio Shack tester, requires test probes or your choice of battery holders and some system of selecting load resistors appropriate for each size of cell. This latter version might be more "accurate" (because you pick the loads), but it would certainly be less convenient.

The audible tester circuit fits on a perforated board measuring 2 by 3-1/2 inches. We had a very small speaker that allowed the audible Radio Shack attachment to fit into a project box measuring 2-1/2 by 5-1/8 by 1-3/4 inches. The Radio Shack tester measures 2 by 5-1/2 inches; when affixing this to our cabinet, there was a bit of overhang at the ends.

For the version carrying the Radio Shack tester, the project box was turned cover-side-down so that the cover could become a removable bottom plate for battery replacement. The Radio Shack tester was mounted piggyback on what used to be the bottom of the project box. The case of the Radio Shack unit has a bottom tray which is held in place by two self-tapping screws (hidden under a paper label. We discarded their bottom portion and used those same screws to secure the tester to our audible box.

Using the Radio Shack tester does not come without a few hitches. For one thing, a 1 by 4 inch rectangle must be cut in the project box to accommodate the spring-loaded slide assembly of the universal battery holder.

Also, as will be described in the circuit, the Radio Shack tester uses separate circuits for running the meter--one for individual cells and the other for 9-volt batteries. We had three choices about how to handle this problem: One was to provide a switch to select where we tapped off the Radio Shack circuit--an easily forgotten switch that would lead to human error. Second, we could have modified our circuit to read the minute voltage across the visual meter; this would have required the use of separate comparator chips, each with offset adjustments. Rather, we chose to modify the Radio Shack Tester so that the 9V and 1.5V circuits work into the same load (50mA on the 9V battery); this simplified our circuit design considerably.

How It Works

A precision triangle wave was created whose positive and negative excursions are symmetrical with respect to a baseline of 1.25V, and whose peak-to-peak value can be calibrated. A comparator (the fourth op-amp in an LM324) has its positive input looking at this triangle and its negative input looking at the cell under test.

If the cell voltage is 1.25V, the comparator's output goes high for half the period of the triangle. If the cell voltage is high--say, 1.5 volts--only the very tops of the triangle will bring the comparator's output high; the comparator will be high for only short durations. If the voltage is below the triangle altogether, the comparator will stay high.

The output of the comparator either stays high all the time (for a very dead cell or for no cell at all), or it produces a pulse waveform whose duty cycle depends on the comparison of the cell voltage with the excursions of the triangle wave. The triangle wave's frequency is 2 or 3 Hertz, depending heavily on where you set its peak-to-peak value (see Calibration), but the duty cycle of the resulting beeps can easily be judged regardless of its frequency.

The output of the comparator operates a beeper using a 555.

Note that by changing the comparator's inputs, the tester would beep solidly for good cells, stay silent for dead ones, and would give forth short pips for near dead ones. Our first experiments were arranged this way until we reasoned that a silent tester could easily be left on after use. An automatic on-off switch would have fixed this; I print that here as an option, but you have to stop a circuit somewhere. (Besides, a mechanical switch is less leaky than an FET switch, and the tester's battery will last the longest with a mechanical one. Of course, one could wind a coil around a magnetic reed switch that would serve as a load and turn the tester on ... Stop! Stop!)

Technical Description and Tester Circuits

An LM336 is used to create a 2.5V standard; a voltage divider from this point establishes a 1.25V reference around which the triangle is based. An LM324 performs the following:

Creation of the triangle requires two op-amps. One is an integrator that produces the slopes; a slope's direction--positive or negative--is directed by a 2N2222. When this transistor grounds an input resistor of the integrator, the slope is positive; when the transistor is open, the slope is negative.

The other op-amp, in conjunction with an enhancement-mode FET, comprises a comparator. (This assembly is called the "composite comparator" in the circuit description.) The FET makes for an output which goes very near 0 and which is clamped to 2.5V. (Note that 1.25V is the midpoint between the output swings.)

One comparator input is looking at the output of the integrator. The other input of this comparator goes to a feedback network which gives it hysteresis. The drain of the FET, which goes between 2.5V and ground, works into a voltage divider (consisting of 120K and a 50K pot). The far end of this divider is fixed at 1.25V, while its tap goes to the plus input of the comparator. The pot is set so that the ramps from the integrator, in order to trip the comparator, have to overcome opposing one-quarter-volt swings imposed by this voltage divider.

A solid version of the 1.25 volts is established by a follower in the LM324. Finally, the fourth op-amp in the package is the main comparator used to test the cell.

The charging resistor of a 555 oscillator is operated by the main comparator.

Circuit for the Radio Shack Battery Tester

The negative side of the meter goes to the spring-loaded negative terminals of the single-cell holder, and to the negative post of the 9-volt test points. The meter's positive terminal goes through 2.2K to the lower positive terminal of the single-cell holder--the terminal used for AAA and N cells. This terminal goes through 20 ohms to the upper positive terminal--the one used for AA, C and D cells; this latter terminal goes through 10 ohms to the spring-loaded negative posts.

Across the 9-volt posts is 950 ohms. The positive post goes through 18K to the positive of the meter.

Modification of the Radio Shack Tester for use with the SK Audible Circuit

The resistors on the 9-volt battery test points are removed and discarded. The 950-ohm shunt is nestled behind the single-cell holder, and the 18K unit is spliced into the wire that leads all the way up to the meter. This wire is removed from the meter terminal and moved down to the lower positive terminal of the single-cell holder; I soldered it to the far lead of the 2.2K resistor. A new resistor, 150 ohms, goes from the bottom end of this wire to the positive 9-volt post. The resultant circuit change can be stated:

The negative side of the 9V battery goes to the negative of the meter as before. The positive 9V terminal goes through 150 ohms, then through the 30-ohm series string back to the negative of the meter. The junction of the 150 ohms and 30-ohm string goes to the lower positive terminal of the single-cell holder, as well as going through 2.2K to the positive of the meter.

Circuit for the Audible Battery Tester

The negative of the 9V battery is grounded. Its positive terminal goes through an SPST on/off switch to the plus 9V line. This 9V line is bypassed to ground by 220uF (negative of this cap at ground).

An LM336 2.5V standard has its "anode" grounded. Its "cathode" goes through 1.6K to the plus 9V line. Its "adjust" lead goes to the arm of a 10K pot. The bottom of this pot goes through a 1N914 diode to ground (cathode of this diode at ground). The top of the pot goes through another 1N914 to the junction of the 1.6K and the LM336 (the cathode of this 914 at the top of the pot).

The junction of the 1.6K resistor and the "cathode" of the LM336 is the 2.5V line.

The 2.5V line goes through two 10K 1% resistors in series to ground. The junction of these resistors is bypassed to ground by 1uF (negative of this cap at ground). The output of this voltage divider is a 1.25V reference.

An LM324 quad op-amp has pin 4 going to plus 9V and pin 11 grounded. One op-amp in this package is used to buffer the 1.25V; pin 3 of the 324 goes to the junction of the 10K resistors, while pins 1 and 2 are tied together and are the buffer's output.

The second op-amp, together with a PN0300M power FET, is used as a comparator. Pin 7 of the 324 goes through 56K to the gate of the FET. The source of the FET is grounded. The drain goes through 1.6K to the 2.5V line.

The triangle-wave generator is made using the third op-amp. Pin 10, the non-inverting input, goes to the unbuffered 1.25V--the junction of the two 10K resistors. Between output and inverting input, pins 8 and 9, is a 0.47uF Mylar capacitor. The inverting input, pin 9, goes through 910K to the 2.5V line. Pin 9 also goes through 470K to the collector of a 2N2222, the emitter of which is grounded.

The base of the 2222 goes through 56K to the drain of the FET--the output of the composite comparator.

The drain of the FET also goes through 120K to the top of a 50K trim pot. The bottom of this pot goes to pin 1 of the 324, the buffered 1.25V. The arm of this pot goes to pin 6 of the 324, the composite comparator's "non-inverting" input. Pin 5, this comparator's "inverting" input, goes to pin 8, the output of the triangle-wave generator.

The fourth op-amp is a simple comparator, with pins 12 and 13 being non-inverting and inverting inputs, respectively. Pin 12 goes to pin 8, the triangle wave output. Pin 13 goes through 56K to ground, and this pin 13 is the positive test lead. The negative test lead is grounded.

A 555 is used as an audio oscillator. Pin 1 is grounded, while pins 4 and 8 go to the plus 9V line. Pins 2 and 6 are tied together and go through 0.01uF to ground.

Pins 2 and 6 go through 56K to pin 7. Pin 7 goes through another 56K resistor to pin 14 of the LM324.

Pin 3 of the 555, its output, goes through a resistor (from 47 to 220 ohms), then through the speaker to the plus 9V line.

Tapping into the Modified Radio Shack Tester

The ground line of the audible tester goes to the negative battery terminals in the Radio Shack unit. Pin 13 of the LM324, the positive test lead, goes to the far side of the 2.2K resistor in series with the meter, this resistor lead going to the "lower positive terminal" on the single-cell holder.

Attachments for the Self-Contained Tester

The negative test lead, which is ground, goes to one side of a male panel-mount connector of your choice (I would use a 2-prong Jones male). The positive test lead, pin 13 of the LM324, goes to the hot side of this connector.

A selection of load resistors should be available for testing various sizes of cells: 6.2 ohms 1-watt for D cells, 15 ohms 1/2-watt for C cells, 30 ohms 1/4-watt for AA cells, 75 ohms 1/4-watt for AAA and N cells, and 150 ohms 1/4-watt for cells such as those in 9-volt batteries. A 9V battery must go through 750 ohms, then through the 150-ohm resistor to ground, with the positive test lead of the instrument looking at the junction of these resistors. (An appropriate load for button cells depends on the sizes being tested. I chose the above loads so as to draw 10% of the amp-hour rating. Perhaps lithium backup cells should be tested with loads of lighter proportion.)

The load resistors may be on a rotary switch; the arm of this switch would be grounded, while each position would go through its respective resistor to the hot test lead. Or, each size of cell holder could carry its own resistor. Each holder in your collection sports a cable-mount female connector matching the male on the tester cabinet. If the load resistors are all on a switch inside the box, don't forget to pot a 750-ohm resistor in series with the 9V battery connector.

Automatic On-Off Switch for the Experimentalist

A P-channel enhancement-mode power FET is used, such as the Siliconix VP0300M. Its source goes to the positive of the 9V battery that runs the audible tester; the negative of this battery is grounded. The drain of this FET goes to the plus 9V line. The gate goes through 4.7 megohms to the source. The gate also goes to the collector of a 2N2222 transistor (must be low leakage). The emitter of the 2222 is grounded. The 2222 base goes through 1.6K to the positive test terminal.


The most important voltage in this device is the buffered 1.25V. The hysteresis voltage divider, being anchored to this voltage, sets the baseline of the triangle wave. Therefore, use a digital voltmeter to set the 10K trim pot associated with the LM336 so that 1.25 volts is present on pin 1 of the LM324. Between you and me, a fair calibration can be done by testing a nicad cell setting the 50K trim pot to the point where the beeper is half-on/half-off. This nicad should be fully charged, but rested for a day, so that its terminal voltage will have settled to 1.25 volts.

Next, the 50K trim pot can be set to suit your liking (depending on what you call "good" and "bad"). Connect a variable power supply to the 9V terminals (the right-hand post is the positive one on the Radio Shack tester). Then, set the power supply for 6 times the cell voltage that you want to consider "dead." For example, set it for 6 volts--one volt per cell in a six-cell string. Once the power supply is set where you want it, adjust the 50K pot so that the beeps just blend together into a solid tone.

The Radio Shack tester has its own definition of "bad"; it has a questionable region between 0.93V and 1.1V. We set our prototype to emit a solid beep with the battery meter reading at the lower edge of the question zone (5.58V on the 9V connector). (That's too optimistic for me; I'd like to start chucking batteries before they drop to 1 volt.)

Notes on Use with Nicads

Nicads work over a small voltage range. Out of the charger, they may show perhaps 1.3V per cell for the first day. After that, they maintain a voltage between 1.2V and 1.25V fairly constantly. After dropping below 1.1V, they require recharging.

Despite what you hear in the street, killing a string of nicads completely is not a good idea. This bears the substantial risk that a robust cell in the string will drive weaker ones backwards (charging the victims in reverse polarity), and this will further undermine the function of the weaker cells. Perhaps deep discharge of a string would be all right if all cells were exactly matched, but they never are.

Matching nicad cells in a battery is always advocated by manufacturers. They will tell you never to replace just a bad cell, but to replace the whole string. With that in mind, it should be noted that testing one cell with a beefy load (like the 10-ohm load in the Radio Shack tester) may put this cell at a disadvantage with respect to its brothers that you don't test. This 10-ohm load would be most significant for an AA cell with very little charge left.

It is probably good manners to methodically test all nicads in a string, taking the same time for each one.

Remember, too, that nicads have a lot of leakage, and they don't hold their charge very well sitting on the shelf. Thus, while cells you charged four months ago may look okay to a tester, they will let you down after a short period of use.

Parts List

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

  • 1--47 ohms to 220 ohms, depending on your speaker
  • 1--150 ohms (for modifying Radio Shack Tester)
  • 2--1.6K
  • 2--10K 1%
  • 5--56K
  • 1--120K
  • 1--470K
  • 1--910K

Load Resistors (for self-contained instrument):

  • 1--6.2 ohms 1-watt (for D cells)
  • 1--15 ohms 1/2-watt (for C cells)
  • 1--30 ohms 1/4-watt (for AA cells)
  • 1--75 ohms 1/4-watt (for AAA and N cells)
  • 1--150 ohms 1/4-watt (the bottom half of the 9-volt divider)
  • 1--750 ohms 1/4-watt (for top half of the 9-volt divider)

Trim Pots (10- or 20-turn):

  • 1--10K
  • 1--50K


  • 1--0.01uF disc ceramic
  • 1--0.47uF Mylar
  • 1--1uF 10V electrolytic 1--220uF 10V electrolytic


  • 2--1N914 silicon diodes
  • 1--2N2222 or 2N2222A NPN silicon transistor
  • 1--Siliconix VN0300M or VN10KM N-channel CMOS power FET
  • 1--LM336 voltage reference
  • 1--555 timer chip
  • 1--LM324 quad op-amp


  • 1--small speaker, Jameco TS30S (use with 47 ohms series resistor)
  • 1--SPST on-off switch
  • 1--cabinet, at least 2-1/2 by 5-1/8 by 1-3/4 inches
  • 1--Radio Shack "handy Checker," No. 22-096
  • 1--9V battery and connector (the only battery tester that needs a battery.)



Most instrumentation projects require standard voltages, and when protoboarding them, I get tired of setting up such a standard each time and calibrating it--for each and every project. The National LM336 is now commonly available and very stable, so I built a box with a bunch of standards in it. It gives me a fixed 1.25V (inspired by the battery tester of this issue), a fixed 2.5V (required by all those pressure gauges and thermometers of past issues), a calibrated variable low-current voltage from about 0.6V to 2.5V, an uncalibrated variable voltage that delivers up to about 2V at 500mA, and a calibrated 0 to 250mA current source (inspired by the ohmmeter in the Talk-&-Tones meter).

I keep a lot of electrical "jigs and fixtures" around. I have little boxes with differential amplifiers in them, little boards with half-built 555 oscillators on them (donning screw-terminal barrier strips for quick connection of frequency-determining components), and a "Transistorized Auditory Gimmick" (see SKTF, Spring 1981) having nothing but sockets and screw terminals so that various component arrangements can be quickly tried. I've got test boxes--I don't even remember what they do any more.

One of these "jigs" is this box of standards. Ideal for reusing a cabinet that housed one's last bad idea, it has a 6-terminal barrier strip for the outputs, two screw terminals for connection to a power supply, and a pointer knob on a precision 10-turn pot.

The pot is fitted with a Braille scale; a very simple one with five divisions. A 3-dot mark is placed at zero and marks every complete turn of the 10-turn pot. The other four marks, equally spaced around the dial, mark every 50mV and every 5mA--the calibrated voltage and current sources both controlled by this pot. (The high-current variable voltage is also controlled by this pot, but the graduations are meaningless, since it uses an emitter follower off the low-current voltage source.)

The board sports an LM336, an LM324 quad op-amp, and two 2N3055 or 2N3054 power transistors which are fitted with separate piggyback heat sinks. Remember that this puppy needs to dissipate some power--nine watts when all sources are working. Therefore, if you put this in an enclosure, pick a real Swiss-cheese item, a cabinet whose project had a lot of miss-drilled holes.

Caution: The current "source" is actually made from a current sink; it is not common to ground, but is based on VCC. Do not simultaneously hook voltage and current sources up to the same circuit.


This requires an external power supply--from 9 to 12 volts. The positive input binding post goes through a 1-amp fuse (I promise to put one in mine right away) to the VCC line. The negative input binding post is grounded. The negative sides of all the voltage sources are common to ground.

An LM336 has its "anode" grounded; its "cathode" goes through 2.2K to VCC. The 336's "adjust" pin goes to the arm of a 10K trim pot. The bottom of this trim pot goes through a 1N914 diode to ground (cathode at ground). The top of this pot goes through another 1N914 to the junction of the 2.2K and the 336 (the cathode of this diode at the pot).

The output of the LM336, its junction with the 2.2K resistor, goes through two 10K 1% resistors to ground. The junction of these resistors is an unbuffered 1.25V point, and is bypassed to ground by 1uF (negative of the cap at ground).

An LM324 has pin 11 grounded and pin 4 going to VCC. Pins 1 and 2 are tied together and go to the fixed 2.5V output terminal. Pin 3 goes to the output of the LM336, its junction with the 2.2K.

Pins 6 and 7 of the 324 are tied together and go to the fixed 1.25V output terminal. Pin 5 goes to the unbuffered 1.25V point, the junction of the 10K resistors.

A ten-turn precision panel-mount pot (value not critical) has its counterclockwise end grounded. Its clockwise end goes to the output of the LM336. Its arm goes to both pins 10 and 12 of the LM324; this arm is bypassed to ground by 10uF (negative at ground).

Pins 13 and 14 of the 324 are tied together and go to the calibrated low-current 2.5V output terminal. This output also goes to the base of a 2N3055 or 2N3054 power transistor.

The collector of this power transistor goes to VCC. Its emitter goes to the high-current uncalibrated variable output.

The emitter of another 2N3055 or 2N3054 goes through a 10-ohm 1-watt resistor to ground; this emitter also goes to pin 9 of the LM324. Pin 8 of the 324 goes through 820 ohms to the base of this transistor. The collector goes to the negative output terminal of the calibrated current source. The positive terminal of this current source goes to VCC.


Offsets in the op-amps will guarantee that the outputs will not exactly agree. Pick your favorite output and calibrate that one. The calibrated variable one may be best, since you can set this to the top of a voltmeter's range where its error is least significant. The 10K trim pot is used for calibration of voltages.

From there on, I juggled around with the emitter resistor of the current source. I didn't have a precision 10-ohm resistor, so I started with two 22-ohm 1/2-watt resistors in parallel, then padded them by adding high-value parallel resistors, starting with a trial 120 ohms until it read what it was supposed to on a current meter.

Remember that the LM324's output can't go to ground, only to within 0.6V of ground, so the first 2-1/2 turns of the calibrated pot must be ignored on the voltage source. This will not be so for the current source, since a 0.6V forward bias on the transistor base must already be accounted for by the op-amp.

Parts List

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

  • 1--10 ohms 1-watt 1% (I concocted this out of two 22 ohm 1/2-watt units in parallel, then padded the assembly until the current source read correctly.)
  • 1--820 ohms
  • 1--2.2K
  • 2--10K 1%


  • 1--10-turn panel-mount precision pot fitted with a Braille scale, 10K, or any value between 500 ohms and 1 megohm
  • 1--10K trim pot


  • 1--1uF 3V electrolytic
  • 1--10uF 3V electrolytic


  • 2--1N914
  • 2--2N3054 or 2N3055 power transistors (need not match)
  • 1--LM336
  • 1--LM324


  • 2--input terminals, binding posts or something
  • 1--6-terminal barrier strip
  • 2--heat sinks for the power transistors (sufficient to dissipate a few watts)
  • 1--1 amp slow-blow fuse with holder
  • 1--open-air container
  • Dozens--projects to come needing references