A Quarterly Publication of
The Smith-Kettlewell Eye Research Institute’s
Rehabilitation Engineering Research Center
William Gerrey, Editor
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 Details regarding products, suppliers, and other contact information are original and may be outdated.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
TABLE OF CONTENTS
by Albert Alden
The auditory oscilloscope is a system which provides an output tone whose pitch (frequency) is an indication of the amplitude at a chosen horizontal position of the signal displayed on a visual oscilloscope. A control determines the location along the horizontal axis at which the signal is sampled. The user can thus identify the waveform and determine its shape, period and amplitude. [Editor's Note: The VCO system is in itself useful, whether or not you build the whole oscilloscope adaptor.]
Al's "System Description," though faultless and complete, goes awfully fast; I'd like to step in here and describe the instrument in physical terms.]
[His original version consisted of an external box connected by a cable to the visual oscilloscope. Protruding from the box was a 4-inch long slide control (a potentiometer which is a bit hard to come by). A pointer on this control could be moved along a strip of Dymo tape which was embossed with braille markings; these markings corresponded with the 10 centimeters of the horizontal axis on the screen.]
[As you scanned the control from side to side, an audible tone "described" the waveform.]
[Later versions differ only in their cosmetic features. For example, it seemed to make little or no difference whether the scanning control was a slide pot or a rotary one; however, a suitable precision slide pot with a long travel distance is perhaps five times the cost of its rotary equivalent.]
[Once the decision was made to use a rotary control, building the auditory attachment inside the oscilloscope became feasible, and it is just such an instrument which I use in my lab. (Mr. Alden used a low-cost Tektronix scope (T921) which was available in both single-trace and double-trace versions; getting the single-trace unit then gave him room to install his adaptation.)]
[Yet another outboard unit was built which contained an "audible meter," so that the vertical amplitude could be measured in centimeters. In this circuit description, a convenient take-off point is indicated for connection to a talking multimeter (or other voltmeter), thus offering this feature as an add-on.]
[Later on, Richard Oehm built an automatic scanning circuit for his auditory scope. He uses the instrument to present the output of a spectrum analyzer audibly; he can then tune complex filter systems while listening to a recurring presentation, thus eliminating the need for scanning by hand.]
[It should be noted that this system can only be used to present single-valued functions. This is true, since the vertical deflection signal cannot be indicated as being in two places at once. Therefore, lissajou patterns and dual-trace presentations are out of the question. (By using his head, a clever technician can set up a switching arrangement with which an A/B comparison of two or more synchronized signals could be made.)]
[How well does it work? Just fine, if you use it sensibly. If you synchronize the scope to a slow version of the clock signal in your digital projects, you can then look at various pulse trains one at a time; it will tell you how often, how high, and exactly where pulses occur. If you look at ramps with the instrument, you will know just where they begin and end, and gross nonlinearities will usually be apparent.]
[On the other hand, assessing and/or minimizing distortion on a sine wave is a bit much to ask, and I should think that using a distortion analyzer would be more appropriate. You can also miss some fine detail such as short spikes and noise--these just don't give the sample-and-hold circuit time to react. (Some scopes provide a way of closely scrutinizing small portions of the display, and with these you could get some of the detail back by going through the waveform with a fine-toothed comb.)]
[Perhaps the most conservative thing I could say about the auditory scope is that it greatly decreases the number of times I have to ask a sighted colleague to come look at a scope for me. For this reason alone, it's a pretty hot item on my list. Bravo, Al!]
The system consists of a modified oscilloscope, a set of electronics with a potentiometer and braille scale for user control, a voltage-controlled oscillator, and a loudspeaker.
The modification of the oscilloscope consists of making connections from the vertical and horizontal deflection signals to the electronics.
In the adaptation, the vertical and horizontal signals are fed into differential amplifiers to provide signals which are referenced to ground potential. The output of the horizontal amplifier and the voltage at the wiper of the horizontal position pot are the inputs to a comparator. When the horizontal signal (a ramp) falls below the position control voltage, the comparator's output rises and triggers a one-shot multivibrator. (Note the use of the ramp with a negative slope.) The one-shot produces a short pulse which activates a sample-and-hold circuit. This circuit samples the vertical signal and stores its amplitude until the next pulse from the one-shot. The next pulse occurs with the next horizontal ramp (sweep).
The effect of moving the position pot is to pick the point along the horizontal axis at which the vertical signal is sampled. (An added frill is to modulate the brightness of the visual display with a differentiated pulse off the comparator's output; this generates a sort of "cursor" by which sighted observers-can see the point at which the auditory system is set.)
The output of the sample-and-hold is fed to the voltage-controlled oscillator. The voltage-controlled oscillator (VCO) includes an exponential converter circuit. This causes equal changes of vertical amplitude on the oscilloscope to produce equal ratios of output frequencies (constant musical intervals). The output of the VCO is fed to a power amplifier in order to drive the loudspeaker.
Universal Hookup Requirements
The "universal" circuit for the auditory oscilloscope can be tailored for use with most general-purpose instruments. The accompanying schematics show the circuitry.
Universal Input System
The interface circuit is used where the vertical and horizontal signals are riding on a large common-mode voltage and/or when any drift of the common-mode voltage may be a problem. This is the situation with older vacuum tube scopes. The circuit values shown were picked for connecting to the deflection plates of a Tektronix Model 561. In this scope, the common-mode voltage is about 180 volts. For other scopes, resistors in the input voltage dividers should be selected to give less than 4 volts bias at pin 3 of the LM310's and a peak signal such that there will be no limiting at the collectors of the 2N2222A's at maximum signal. The maximum signal occurs when the peak-to-peak vertical deflection just fills the face of the CRT.
Another resistor is selected so that the maximum vertical signal at the top of its associated 10K pot is less than 5V.
The signal from the wiper of the above pot is buffered with an LM310 follower; then it is fed into a system of analog switches (contained in a CD4066). These switches serve two purposes: The first is to gate the vertical signal through to a sampling capacitor. The second is to select the most appropriate sampling capacitor for a given setting of the sweep rate control.
Two sampling capacitors are available, 100pF and 0.01uF. Selection of the appropriate one is done by controlling the CD4066 with a mechanical switch (this could be a toggle--in our self-contained scope, we used an extra pole on the sweep-rate control to apply the-appropriate signal to the 4066).
Two more resistors in series with the scanning control must be selected to allow setting of the horizontal position pot to operate the LM311 comparator over the full range of the horizontal signal. The two 10K variable resistors are adjusted to calibrate the movement of the position pot with the graticule marks on the face of the CRT. In other words, the selectable resistors are "coarse" adjustments, while the 10K trimmers are "fine" adjustments. Therefore, before selecting the resistors, set the trimmers to mid-position so that the eventual fine calibration can be accomplished.
If the interface circuit is not used, a voltage divider may be needed between the horizontal signal source and the 15K resistor to attenuate the signal to within a 0 to plus 12V range.
Power Supply Requirements
The circuit is designed to operate from plus 15 VDC, plus 5 VDC, and minus 15 VDC. The plus 5V may be derived from the plus 15V by using the LM3405 as shown on the sample-and-hold schematic.
Universal Interface Circuits
Two carbon copies of this circuit must be built--one for the vertical deflection plates, and another for the horizontal ones. They each consist of an LM310 follower and three transistors
The plus deflection plate (or other differential signal source) goes through a 510K resistor, then through a "selected resistor" to ground. The junction of these two resistors goes to the input (pin 3) of an LM310 follower. The output, pin 6, goes to the base of a 2N2222A. The collector of this transistor goes through 2.2K to plus 15V. This collector becomes the "negative signal output" ; this is true because of the inversion that takes place by way of the transistor.
The negative deflection plate goes through 510K, then through an identical "selected resistor" to ground, with the junction of these resistors going to the input of another LM310 (pin 3). The output of this follower (pin 6) goes to the base of another 2N2222A.
The collector of this transistor goes through 2.2K to plus 15V; this becomes the positive output of the differential interface circuit.
Each of the above transistors has an emitter resistor of 510 ohms; the far ends of these are-tied together and go to the collector of a third 2N2222A. The emitter of this third transistor goes through 270 ohms to minus 15V. Its base goes through 3eK to ground, as well as going through 10K to minus 15V.
To power the LM310's, their pin 7 goes to plus 15V, while their pin 4 goes to minus 15V.
As mentioned before, the selectable resistors were chosen for a Tektronix 561, and turned out to be 10K. They should be selected to give less than a 4V bias at the inputs of the LM310's, and so that the peak signal does not cause limiting at the collectors of the 2N2222A's.
Sample and Hold Circuit
The "positive vertical output" is used; this comes from the collector of the transistor whose base is driven by the buffered signal from the negative vertical deflection plate. This collector goes through a "selected resistor" (36K in our unit) to the top of a 10K pot (this is the scaling adjust for the VCO), the bottom of which is grounded. This resistor is selected so that the maximum peak signal at the top of the pot is less than 5V.
The wiper of the 10K pot goes to the input (pin 3) of another LM310, with the output of this 310 (pin 6) going to pins 3 and 2 of a CD4066 analog switch. Pins 2 and 3 of the 4066 are each one terminal of an SPST switch; the other terminals of these switches, pins 1 and 4, are tied together and go through 100pF to ground. (In other words, these two switches are in parallel. As will be evident, one is always triggered with a short pulse from a one-shot, while the other receives longer pulses when appropriate.)
Pins 1 and 4 not only go through the 100pF unit to ground, but they also go through 0.01uF to pin 8 of the 4066 (the terminal of another switch). Pin 9 of the 4066 (the other side of this switch) is grounded. Pin 14 of the 4066 goes to plus 5V, and pins 7 and 11 are grounded.
Pins 1 and 4 of the 4066 also go to pin 3, a non-inverting input, of an op-amp (one of two contained in a CA3240). Pins 1 and 2 (output and inverting input) are tied together to make a follower. Pin 1 is the output of the sample-and-hold system which drives the VCO.
The second half of the 3240 is not used (pin 5 is non-inverting input, pin 6 is the inverting input, and pin 7 is the output). This makes a convenient amplifier from which a voltmeter could be driven; its gain could be chosen so as to make the scale of the meter come out in centimeters. Not only must the amplifier gain be adjustable to get proper scaling for the meter, but provision must be made for taking care of an offset voltage. Although untried, the following is a suggested circuit:
Pin 6, the inverting input, goes through 20K to the arm of a 10K pot, the top of which goes to plus 5V and the bottom of which is grounded. Between pins 6 and 7 is a 39K feedback resistor. Pin 5 goes to the arm of another 10K pot, the bottom of which is grounded; the top of this pot goes to pin 1 of the same 3240. The first pot adjusts the offset, and the second pot the scale factor. There will be interaction between these two adjustments. The voltmeter can then be connected between pin 7, the output of this opamp, and ground.
To power the LM310, pin 7 goes to plus 15V, while pin 4 goes to minus 15V. On the CA3240, pin 8 goes to plus 15V, while pin 4 goes to minus 15V.
The negative horizontal signal is used to trigger the system; this is gotten from the collector of the transistor whose base gets the buffered version of the positive horizontal deflection plate. This collector goes through 15K to the plus input of an LM311 comparator (pin 2). From pin 7 to pin 2 (from output to non-inverting input) is a 1.5meg resistor. The 311 output requires a pull-up resistor; pin 7 goes through 1K to plus 5V. This output also needs a negative reference; pin 1 goes to ground. (Getting the output to operate between ground and 5V is essential, since the comparator triggers a TTL chip.)
To power the 311, pin 8 goes to plus 15V, while pin 4 goes to minus 15V.
The negative input, pin 3 of the 311, goes to the wiper of the scanning control. (This is a 10K linear precision unit.) The top of this pot goes through a "selected resistor" (20K in our unit), then through a 10K trimmer rheostat to plus 15V. The bottom of the pot goes through another "selected resistor" (20K in our unit), then through another 10K trimmer rheostat to minus 15V.
The output, pin 7, of the LM311 can be used to modulate the intensity of the display by going through 15pF to the Z-axis input. (Note: This works for scopes which require a negative-going signal for an increase in intensity.) This output also goes to pins 1 and 9 of a 74LS123; these are the inputs of two one-shots. Of the first section, pins 2 and 3 are connected to plus 5V. Between pins 14 and 15 is connected a 100pF capacitor, while pin 15 goes through a variable resistor to plus 5V. It is adjusted to generate a 200-nanosecond pulse from this one-shot.
Pin 13 of the 74LS123 (its output) goes to pin 13 of the CD4066 (one of the switch's control pins).
On the second half of the 74LS123 (whose pin 9 input already goes to the comparator), there is a capacitor of 0.001uF connected between pins 6 and 7, with pin 7 also going through 27K to plus 5V. Pin 11 goes to plus 5V. The output, pin 5, goes to the control pin of the other parallel-connected switch (pin 5 of the 4066).
Pin 10 (the enable of the second one-shot) of the 74LS123 goes to the arm of a single-pole double-throw switch (sweep-rate selector). Along with the enable of the one-shot, the control of the third switch (pin 6 of the 4066) also goes to the arm of this selector switch. The "fast" position of the SPDT switch is grounded, while the "slow" position goes to plus 5V.
Pin 16 of the 74LS123 goes to plus 5V, while pin 8 is grounded. This plus 5V source comes from an LM340-5 three-terminal regulator. Its common terminal is grounded, while its input terminal goes to plus 15V. Between input and common is 0.1uF; the output terminal goes through 10uF to ground (negative of the capacitor at ground).
Exponential converter, Voltage-Controlled Oscillator, and Speaker Amplifier
One half of a CD3240 is used as an inverting amplifier with offset; this is in order to set the "base-line pitch" of the VCO system. The second half has a diode in its gain-determining circuit; the non-linearity of the diode creates an amplifier gain which is approximately exponential.
Pin 4 of the exponential converter's IC goes to minus 15V, while its pin 8 goes to plus 15V.
Pin 3 of this IC, a non-inverting input, is grounded. The inverting input, pin 2, goes through 10K back to pin 1 of the sample-and-hold follower (an aforementioned 3240). Pin 2 also goes through 3.9K to the wiper of a 10K pot (this is the offset adjust for the VCO). The top of this pot goes to plus 5V, while the bottom end is grounded. Between pins 1 and 2 (the output and inverting input) is a 1.3K feedback resistor.
The output of the above stage, pin 1, goes to the cathode of a diode, the anode of which goes to the inverting input (pin 6) of the other op-amp in the chip. (The diode is a 1N4148 silicon unit.) Pin 5, the non-inverting input, is grounded. The feedback resistor from pin 7 to pin 6 is 47K.
Both sections of another CD3240 comprise the oscillator. In this case, the 3240 is operated from 5V; pin 8 goes to plus 5V, while pin 4 is grounded.
The output of the previous stage, its pin 7, goes through a 100K, 1% resistor to pin 2 of this 3240 (an inverting input). Between pins 1 and 2 (output and inverting input) is a capacitor of 820pF. Pin 7 of the previous IC also goes through a 49.9K, 1% resistor to pin 3, the non-inverting input, of the oscillator's 324). A second 49.9K, 1% resistor goes from this pin 3 to ground. The inverting input, pin 2, also goes through another 49.9K, 1% resistor to the collector of a transistor (2N2222A). The emitter is grounded.
Pin 1, the output of the first half of the 3240, goes to pin 6, the inverting input, of the same package. The non- inverting input, pin 5, goes through 39K to the junction on a voltage divider--two 4.7K resistors in series between plus 5V and ground. Between pin 5 and the output of its op-amp, pin 7, is another 39K resistor. Pin 7 also goes through 7.5K to the base of the 2N2222A.
The output of the oscillator is actually pin 1 of its op-amp. This pin 1 goes to the positive end of a 1uF coupling capacitor, with the negative end going through 39K to the top of a 10K volume control. The bottom of this control is grounded.
The wiper of the control goes to pin 3 of an LM386. Pins 2 and 4 of the 386 are grounded. Pin 6 goes to plus 5V. Pin 5, the 386's output, goes through 10 ohms in series with 0.068uF to ground. Pin 5 also goes to the positive end of a 100uF capacitor, the negative end of which goes through the speaker to ground.
[The editor would like to draw your attention to the oscillator circuit. The first half of the 3240 is an integrator which is directed to integrate either upward or downward as orchestrated by the 2N2222A transistor. A comparator looking at the output of the integrator turns the transistor on, then off, then on again; a triangle wave out of the integrator is the result. This comparator has considerable hysteresis, given its positive feedback network. This hysteresis determines the amplitude of the oscillations; increasing the feedback resistor will decrease the amplitude.]
[This whole section, the exponential converter, the oscillator, and the audio amp, is worth building for other uses. Most "gimmick circuits" do not account for the fact that you cannot hear changes in pitch at high frequencies as well as you can at low ones. This circuit does account for that fact, since the exponential converter keeps the ratio constant for a given change in input signal; i.e., the pitch change is greater at high frequencies than at low ones. Try it, you'll like it.]
First of all, the information for choosing the selected resistors is gathered together and presented here. After this, calibration procedures for three sections will be described: the gain of the sample-and-hold system and the VCO's offset are set. Then, the horizontal scanning control is calibrated so that its braille markings correspond with the graticule of the screen. Finally, the external vertical output's offset and gain are adjusted for use with an external meter (this section is optional).
If the signals are to be gotten from the deflection plates, resistors in the initial voltage dividers (the ones going to ground off the 510K units) must be chosen for two criteria. First, the baseline deflection should create a bias at the LM310 inputs which does not exceed 4 volts. Second, make sure that with deflections whose peaks fill the screen, no limiting occurs at the output collectors. You can check these criteria as follows:
Arrange a horizontal line to be present in the center of the screen. (Do this by first causing the horizontal sweep circuit to trigger on something--the ac line for example; then, with the input shorted, set the vertical position control to place this line in the center.) Now, pick the resistors (the ones immediately following the 510K units) so that less than 4V appears at pin 3 of the LM310's. Next, move the horizontal line from bottom to top of the screen (using the vertical position control), and with a voltmeter connected to one of the output collectors, verify that no "bottoming out," or "limiting," occurs as these extreme positions are approached. If limiting does occur, reduce the value of these resistors until this ceases to be a problem.
Without another oscilloscope, equivalent tests are harder to make for the horizontal circuit. In most cases, however, the voltages seen by the input circuits will be identical in character; proper choice of the vertical components will give you sufficient information to install these in the horizontal circuitry. Different resistors must be chosen where the screen is not square. For example, if the screen is only 8 centimeters high, but is 13 centimeters wide, the "selected resistors" in the horizontal circuit will have to be reduced by 20% or so.
A resistor at the top of the 10K scaling potentiometer (at the input of the Sample-and-Hold Circuit) must now be chosen so that the voltage at the top of this control does not exceed 5V. With a voltmeter at the top of this pot (the negative meter lead being grounded) and with the horizontal line still being generated on the scope, chose the resistor so that, as the line is moved from top to bottom of the screen, a maximum voltage of 5V is not exceeded.
It should be restated that if the differential input system is not used (and hence, appropriate voltage levels at the input to the horizontal circuit cannot be assumed), steps should be taken to assure that the horizontal ramp signal fed to the LM311 comparator is restricted to a plus/minus 12V range or less. This can most easily be checked by viewing the ramp signal with another oscilloscope. If another scope is not available, slow the sweep down to perhaps 0.5 seconds per centimeter and watch the ramp on an analog voltmeter. In any case, a signal whose peak exceeds the plus/minus 12V limit must be attenuated with a voltage divider before it is fed to the 15K input resistor.
Finally, resistors at the ends of the 10K scanning potentiometer must be selected. Determine the voltage swing which the ramp applies at the 15K-input resistor (this resistor comes off pin 2 of the LM311 comparator). (This can best be done with another oscilloscope, but it can also be done with an analog voltmeter if the sweep is slow enough.) Then, with the 10K trimmers (fine adjustments) set to their midpositions, choose these resistors so that the wiper of the scanning control (and hence pin 3 of the LM311) covers this same range of voltages.
VCO Offset and Scaling
The VCO can now be "calibrated" so that its frequency excursions fall within a comfortable range of audio frequencies. Ideally, the center of the screen should produce a tone of about 530Hz; then, for our units, the arbitrary scale of an octave per 2 centimeters was chosen. For a screen whose height is 8cm, the VCO will span 4 octaves--from 125Hz to 2000Hz. (The exponential converter is not perfect. Depending on the characteristics of your diode, "tracking" may be better in one direction than the other.)
With a horizontal line in the center of the screen, set the "offset" potentiometer (off pin 2 of the 3240 in the Exponential Converter) for a tone of about 500Hz (about 4th octave B). Then, move the line up and down to get a rough idea of where the scaling is at this point; if the oscillator moves wildly, decrease the setting of the pot on the input of the Sample-and-Hold system. When you change the scaling, a shift in the offset will result; these two adjustments profoundly interact. Once again, put the line. in the center of the screen and set the offset for about 500Hz--then check the sealing. After several iterations, you should be able to set up a condition where the oscillator drops about an octave as the line is brought down by 2cm, and goes up an octave when the line is brought to 2cm above the center. [Editor's Note: Remember that this is a relative indicator, not a church organ. Don't frustrate yourself; exact measurements of amplitude can be done without an oscilloscope, just using a voltmeter and a diode detector.]
Horizontal Axis Calibration
A periodic waveform with distinctive features is best suited for this purpose--a squarewave, a sawtooth, or even a triangle. Synchronizing the sweep externally is also desirable, since this will afford maximum flexibility in placing the waveform on the screen. Arrange the repetition rate of the wave, or the horizontal scaling on the scope, to put the distinctive features in coincidence with marks on the graticule.
The trimmers in series with the scanning control can now be adjusted so that the corresponding markings on the tactile scale line up with those on the graticule. Once again, these two trimmer adjustments interact; start out by lining up a mark near the left end, then line up an appropriate mark near the right end. Repeat this procedure until you are satisfied with the results.
If the waveform being used has sharp discontinuities, you will notice erratic behavior of the VCO at these points. Rather than trying to detect the exact center of the discontinuity (which is undefined), strive to find a point of consistency that you can repeatedly find on the waveform, and use this feature for calibration. Whether boundaries of discontinuity are poorly defined or not depends on the character of the signal source, the presence or absence of noise or jitter on the signal, etc. Often, you will notice that some sweep rates and signal frequencies work better than others; it is always worth trying different combinations. [The editor suggests you start out with a signal whose frequency is perhaps 10kHz, and with a sweep rate of 50 microseconds per centimeter.]
External Offset and Gain
The offset desired will depend on what kind of meter you wish to use. For example, if an analog meter is to be used, you may want "center screen" to be represented by "center scale" on the meter. On the other hand, if a digital meter which is capable of reading both negative and positive voltages is used, it would make sense to eliminate any offset, thus putting "center screen" at zero volts.
Scaling of the output signal can be done so as to make sense to the user. For example, if the range of your voltmeter permits, 1V/cm would be ideal.
Set up a horizontal line in the center of the screen, and set the offset as desired (the adjustments interact, so consider this adjustment to be temporary). Note the "rate of change" as you move the line up the screen; then make an appropriate adjustment with the gain control and try again. Repeat this procedure until your criteria are satisfied. (Note, any attempt at this point to change the level control at the input of the Sample-and-Hold circuit will directly affect both gain and offset of this amplifier stage.)
Notes on Using the Scope
The larger sampling capacitor is necessary at slow sweep rates because the "samples" are taken less frequently. In other words, at faster sweep rates, the small sampling capacitor is refreshed often enough so that "sagging" of the held voltage between sweeps is not significant. At slow sweep rates, however, the VCO would drop in pitch between sweeps; the result would be noise (flutter) in accordance with the sweep rate superimposed on the tone. Therefore, the large sampling capacitor should be switched in for sweep rates below 1 millisecond per division.
At very slow sweep rates (10 milliseconds per division or so), the rate of refreshment will be quite apparent as you scan past a feature with a steep slope. On a sinewave, for example, the "slope" (rate of change) is maximum at the zero crossings; in scanning past these points, the VCO will jump in steps. These "steps" will be smaller as you scan slower, since the differences in sampling will be less. If scanning is done steadily enough, the intervals of the steps can be an actual indication as to the degree of slope.
Causing the sweep circuit to trigger on the waveform is very easy with the auditory scope. If the sweep circuit is triggering on something unrelated, to the input signal, the VCD will make warbles, random music, or rough noise. As soon as proper triggering is established, a pure tone is the result. (Of course, this assumes several essential conditions: the tone will be as pure as the signal is free of noise, and the amplitude of the signal must ,be within the workable range of the VCO. If no definite indication of triggering is heard, these conditions should be sought after.)
[The editor has dreamed up this circuit without trying it; watch out for bones.]
This circuit is very fancy; not only is its scanning rate variable, but you can set it to scan to the right, to the left, or back and forth. It uses an ICL8038 function-generator chip, the triangle output of which goes through an amplifier with adjustable gain and offset. It simply goes in the place of the hand-operated scanning control; i.e., pin 3 of the LM311 goes to the output of this circuit instead of to the wiper of the control. By measurement or by experiment, the ramp of this circuit must be made to match the voltage range of the control.
Automatic Scanning Circuit
Pin 11 goes to minus 15V, while pin 6 goes to plus 15V. Pin 7 goes through a 250K pot to plus 15V, with the arm of this pot going to pin 8 (this is the "scan-rate" control, and its wiper is operating the VCO input of the 8038). Pin 4 and pin 5 each goes through a 560K, then through a 1K to plus 15V. Pin 10 goes through 2.2uF to minus 15V (a tantalum unit with its positive end toward pin 10).
A double-pole double-throw switch, with a "center off" position, selects the direction of sweep. On pole A, the "scan up" position goes to pin 4. On pole B, the "scan down" position goes to pin 5. The arm of pole A goes to the junction of the series resistors off pin 4, while the arm of pole B goes to the junction of the series resistors off pin 5.
Pin 3, the "triangle/ramp output," goes to a 10K potentiometer (gain control). The other end goes to ground, and the wiper goes through a 47K to pin 2 of a 741 op-amp. From pin 2 to pin 6, there is a 56K feedback resistor. Also from pin 2, there is a 100K resistor to the arm of a 20K pot (offset adjust). The ends of this pot are connected to plus 15V and minus 15V. Pin 3 is grounded, pin 4 goes to minus 15V, and pin 7 to plus 15V.
Pin 6 of the 741 is the output of this circuit, and it goes to pin 3 of the LM311, which has been disconnected from the scanning control.
These adjustments will not interact, and you will just have to "cut and try" until this ramp voltage matches the available voltage swing from the hand-operated control. With the scan-rate control turned down to a slow speed, the output of this ramp will easily be measurable with a voltmeter.
Resistors (1/4 watt, 5%)
- 1--10 ohm
- 1--270 ohm
- 2--510 ohm
- 1--15 K
Resistors (1/4 watt, 1%)
- 1--10K, 1% linearity
- 1--10K, volume control
Capacitors (disk ceramic)
- 1-100uF, 10V
- 1--10uF, 25V
- 1--1uF, 25V
- 1--SPDT switch
First, "logic probes" are discussed in general terms. Second, a couple of alternative logic-tracing tips are given. Finally, two commercial audible logic probes are surveyed: the Micronta "Digital Logic Probe" marketed by Radio Shack (Cat. No. 22302), and the Production Devices Audio/ Visual Logic Probe," Model 110, marketed by Jensen Tool Company (Cat. No. 138B110).
When a digital project is sitting there, whirring away--doing the wrong thing (practicing "sophomoric logic," as one of my supervisors used to say)--what do you test it with? The answer is, "with an instrument of its own kind"; a detector of "1's" and "0's." A voltmeter and an oscilloscope would do as well, but you would spend more time taking readings and fiddling with the test instruments than is needed to answer the three basic questions: Is there a "0"? Is there a "1"? Is there "activity" whose extremes are "1's" or "0's" (pulse trains and the like)?
The simplest logic probe is a lamp (an LED, for example). The cathode of the LED is grounded, while its anode goes through perhaps 510 ohms to the probe tip. If it lights, the test point is at a logic "1". If it does not, you have a logic "0." You ask, "But what if the test point is an open circuit; how would you know?" This necessity is the mother of invention which brings us to Model II.
Two lamps are used this time--one green and one red. The red one is connected as before; the green one has its anode going to the plus supply, with its cathode going through 510 ohms to the test probe. When the probe is unconnected, they light each other up at half brightness; however, when a "1" or "0" clearly appears at the probe tip, the lamp which stays lit tells the story.
The above circuits have a serious disadvantage. They draw considerable power from the logic circuits under test, and they may very well change the function of the circuit as you test it. (The above system would not work at all for CMOS logic, where the outputs of devices are only capable of sourcing a quarter of a milliamp.) Model III contains a buffer system using either a digital logic inverter or the complementary pair of transistors to follow:
The collector of an NPN transistor goes to VCC; the collector of a complementary PNP transistor is grounded. Between the two emitters is a 22-ohm resistor. The bases are tied together and go to the test probe. The PNP emitter goes through 510 ohms to the cathode of the green lamp, with the anode of this lamp going to VCC. The NPN emitter goes through 510 ohms to the anode of the red lamp, with the cathode of this lamp being grounded.
Now we have a high-impedance probe which will indicate both "1's" and "0's," and it is more sophisticated than the simple "1's" detector provided for us in the digital labs at the university. Trouble arises, however, if we need to know about two other kinds of signals: First of all, an occasional pulse may come by which is too short to see. Secondly, unless a "pulse train" (a digital waveform) is below the flicker rate of vision, which is about 25 Hz, both lamps will light and no indication of dynamic activity will be present.
What we need is a "pulse stretcher" (nothing akin to the mythical "smoke shifter" which tender-foot campers are often sent in search of). A one-shot (a device that, when triggered, stays "on" for a prescribed length of time) can keep the lamp lit long enough to be seen in the event of the occasional pulse.
However, unless we pull another trick, a "pulse train" would keep retriggering a single one-shot; the "1's" lamp would remain on to indicate a "1" state, which is not necessarily the case. A second one-shot can be arranged to hold the first one back for a prescribed length of time between accepted trigger pulses. In this way, a fast pulse train would be indicated as a flicker of the lamps. (By now, I suppose we can call this Model V.)
This last model tells us nothing about how fast (at what frequency) are the pulse trains that it sees; neither does it tell us anything about the duration of an occasional pulse. It does, pretty much, show the user the "character" of information on the various test points being examined. Most importantly, it does so in a hurry! This is worth its weight in gold when you are faced with the 300-plus pins inevitably present in a project containing 20 chips.
There is one more element of good practice which will be part of any well-designed probe--diode protection (and an appropriate current-limiting resistance) on its input. Transients, static charges, or operator error can damage a probe by taking its input "beyond the rails" (beyond the limits of its supply). In fact, any device (be it analog or digital) which has direct-coupled inputs that are subject to the abuses of the outside world should be so protected. In our complementary-pair example, the transistor bases will not go to the probe tip directly, but will be connected as follows:
These bases go to the cathode of a diode (1N914), with the anode of this diode being grounded. The bases also go to the anode of another diode, with the cathode of this unit going to VCC. Finally, the probe tip goes through a resistor (perhaps 1K) to the junction of the bases and the diodes.
If CMOS gates (or other logic) were being used, their input terminal should also go through a resistor (perhaps 100K) to ground. This is true, since allowing a CMOS device to float can cause the channels of the FET's to dissipate more power than they were designed for.
Audible Logic Detectors
Now, all that we need for you and me is a logic probe that beeps instead of blinks. Such a device would give us two clear advantages over the sighted user: First, we would not have to fracture our spines positioning ourselves to see the lights. Second, the occasional pulse would cause an alarm to go off; we would not have to be focused on the lamps at the time of the pulse.
Gee, too bad--a couple of "normals" have already taken note of these obvious advantages, and two (and perhaps more) audible probes exist commercially. One thing they apparently have not thought of, however, is that our "flicker rate" exceeds 12kHz and does not merge everything together in a blur at 25 cycles per second. The discussion to follow addresses these talents.
Listening to Digital Signals
With just a simple test amplifier, digital signals in the audible range can be listened to directly. (In fact, signals well below the audible range can also be heard; remember that the transition of states occurs so rapidly that very pronounced clicks will be heard through the amplifier.) Furthermore, you can tell something about the duty cycle of a pulse train by the timbre of the sound. A squarewave will be a clarinet-like tone. A wave with more even harmonics will be brassier sounding. A train of very narrow pulses will have very little of its fundamental frequency component present (it will give off a "Bizzzzzz").
Connect your probe tip through perhaps 150K, then through 4.7K to ground. The cold input of the amplifier is grounded, while its hot input goes through 0.1uF to the junction of the resistors.
The amplifier is then presented with a highly attenuated version of the signal, appearing across the 4.7K resistor. (Of course, the appropriate choice of values for this voltage divider will depend on the gain of your amplifier--and the logic levels. The above values will keep the attenuated excursions under 0.5V; this is plenty of signal for the standard "high-level input.")
Mr. Noel Runyan, formerly at TSI and now in business for himself, once told me that he got more use out of a series of dividers (flip-flops) than he did from any other fancy instrument, including an oscilloscope. It takes no great theoretician to understand why.
The frequency range over which the ear can reliably judge relative pitches exceeds a ratio of 100 to 1. Having a chain of twelve dividers (as will be shown in a circuit to follow) could conceivably permit frequencies of 17mHz to be identified as musical tones; the range of 100 to 1 is extended to 409,600 to 1. (Sorry, the maximum clock signal my chip will withstand is 2mHz, thus dividing this signal down to 488.3Hz, about 5-1/2 cycles shy of fourth-octave B.)
After a signal has gone through a divider, all information as to its duty cycle is lost--they're all squarewaves at that point. What you know for certain is that the signal is at logic levels, and is sufficient to trigger the dividers. (This latter is something you cannot tell by listening through a simple amplifier, and I have been known to "test" a signal's operation of a divider or a gate in order to affirm its levels.)
The CD4040 contains twelve dividers in cascade, all outputs of which are available for listening to. It triggers on the negative edge of its clock pulse; the negative-going "transition time" must be less than 5 microseconds. The maximum clock frequency is 2mHz for 5V logic, and 6mHz for a supply of
- 16--VDD (plus supply)
- 11--Reset, high for reset (grounding it makes the counter work)
- 9--divides by 2
- 7--divides by 4
- 6--divides by 8
- 5--divides by 16
- 3--divides by 32
- 2--divides by 64
- 4--divides by 128
- 13--divides by 256
- 12--divides by 512
- 14--divides by 1024
- 15--divides by 2048
- 1--divides by 4096
Connect the outputs to a 12-position switch in the order presented above [I didn't know chip engineers used such strong stuff]. The arm of this switch goes through 330K, then through a 10K volume control to ground. The arm of this control then goes through 0.1uF to the input of your amplifier.
Pin 8 of the 4040 is grounded (the ground of the project under test). Pin 16 goes to the plus supply of the project under test.
Between pins 8 and 16 (located close to the chip) is 0.1uF disc. The reset, pin 11, is also grounded.
Pin 10, the clock input, goes through 100K to ground. This pin 10 also goes through 1K to the probe tip. Pin 10 also goes to the anode of a diode (1N914, 1N4148, etc.), with the cathode of this diode going to pin 16. Pin 10 goes to the cathode of a similar diode, with the anode of this unit going to pin 8.
Detection of the RF Signal
A handy device for tuning your clock to a precise frequency is a good general-coverage receiver. I use my Sony ICF2001 portable all-wave radio rather often for this purpose; it tunes as low as 150kHz, which has covered several clock frequencies of interest (that of the notch filter and the audible strobe tuner-both found in SKTF, Summer 1983). Be careful of this trick, however; digital projects contain various signal frequencies which are rich in harmonics. Make sure you know what you're detecting.
Audible Logic Probes
Before the two commercial units are discussed, it should be mentioned that very simple buzzers can do well enough for many applications. For example, Noel Runyan has told me that his favorite instrument for troubleshooting RS232 ports is the Mallory "Sonalert." (The SC628 is the one you want.) It draws very little current (less than 5mA by itself, and less than 0.2mA with 47K in series with it). Being a resonant device, it "rings"; its "off" response is not very fast. Nevertheless, Mr. Runyan, who very much knows what he's doing, says it still presents most of the essential information.
Another approach is to use one of the solid-state buzzers which have a control terminal. An exemplary circuit, using the star Micronics CMB series, is given below:
The CMB-06 (3 to 7 volts) or the CMB-12 (7 to 17 volts) are used, chosen on the basis of the logic levels being detected. Pin 1 is grounded to the project, while pin 14 goes to the VDD line of the project. Between pins 1 and 14 (located close to the buzzer) is the parallel combination of 0.1uF disc, and 10uF electrolytic (negative toward pin 1). Pin 8, the control terminal, goes to the anode of a diode, the cathode of which goes to pin 14. Pin 8 also goes to the cathode of another diode, the anode of which goes to pin 1. Pin 8 also goes through 1K to the probe.
The above easy solutions are simple "1's" detectors, although slower pulse trains will make themselves known by "dirtying up their sound a little. They are just about all that's needed if your tests can be performed with the project in the "static" condition (with the clock stopped).
Commercial Audible Logic Probes
Circuit diagrams are not printed here for the following reasons: In order to print those of the probes listed, I would have to get them by "reverse engineering" (laboriously tracing the PC boards). Second, I don't think you could beat the price much by building your own. Finally, I have not seen a circuit for one which works as well as these--with pulse stretchers (sometimes called "pulse catchers") and the whole shootin' match.
I am interested in running a good circuit, especially for our overseas readers for whom building their own would probably be cheaper. Please keep on the lookout for such a circuit; be very specific about where it came from, since I could get into a lot of trouble.
The units described present "0" and "1" as distinctly different tones. Both have "pulse stretchers" (sometimes called "pulse catchers"), and have some way of telling you about pulse trains. Both can be used with TTL or CMOS logic. (The ranges of acceptable "0" and "1" are different for each.) Both are powered from the circuit under test.
The Micronta "Digital Logic Probe" Radio Shack No. 22-302
This unit is very compact and lightweight, although it is made with discrete components. (Transistors ... twelve of 'em.) It costs about $20.
It has LED's of three different colors, as well as making corresponding tones of three pitches:
- "1"--red, high-pitched tone (1200Hz)
- "0"--green, low-pitched tone (400Hz)
- Pulse--yellow, medium tone (1000Hz)
A normal/pulse switch selects between detection of DC ("static") logic levels and pulse detection. This may be a disadvantage; I have not used the probe enough to know whether or not it tells you when to look for a pulse (I suspect not).
Another switch selects the type of logic being detected; it is labeled TT/LS in one position (meaning it is right for straight TTL, or the low-power Schotky stuff), and CMOS.
Like all active logic probes, it has a cord with alligator clips by which you supply power to the device, which can be done from the supply of the circuit under test. It contains a diode to protect the probe from being hooked up with these leads reversed. However, for some reason, this diode is in the negative power lead; they caution that an error in logic detection (especially with TTL) will result from the drop of this diode unless you use a short plug-in clip lead-grounding this also. But, this defeats the polarity-protection diode, and they therefore suggest that you can use a separate supply. (Humm-mm. This second supply had better be of the same voltage. Better yet, hook it up to the project's supply and test to see if it is correct by looking for a logic "1"; if it is, ground the extra lead and forget their diode.)
Supply Requirements: Operates from 4.7V to 18V. At 5V, the standby current is nominally 20mA; it is 40mA at 15V. Maximum operating current drain at 5V is 45mA, and is 150mA at 15V. (You'd better make sure that the power supply is good for this additional load.)
Input Specifications: 100K input impedance. Maximum input frequency is 10mHz. Minimum detectable pulse width is 50 nanoseconds.
TTL/LS Detection Levels: A "0" is less than 0.8V, plus/minus 0.2V. A "1" is 2.2V, plus/minus 0.2V.
CMOS Detection Levels: Nominally, a "0" is less than 30% of VDD, and a "1" is greater than 70% of VDD. More specifically: with a supply of 5V, "0" is 1.5V, plus/minus 0.3V; "1" is 3.5V, plus/minus 0.4V. At 15V, "0" is 4.5V, plus/minus 0.75V; "1" is 10.5V, plus/minus 0.75V.
The Production Devices Model 110 "Audio/Visual Logic Probe" Jensen Tool Company, 138B110
Costing about $42, this is a very nice instrument (although it is much heavier, 4oz, than the Radio Shack unit above; this is due in part to its using a standard PM loudspeaker.) It is considerably louder, and may be the instrument of choice if your lab has noisy fans going.
Its PC board contains three chips and one transistor. Two of the chips are in sockets; presumably if you blow it up, you stand a chance of fixing it. (When we blew ours up, looking at one set of levels and powering it from another, the soldered-in protection diodes went; the handy plug-in chips did not.
It only has two LED's, red for "1" and green for "0." They describe its output signal for pulse detection as a "warbling, or blended tone." This may mean that the LED's won't adequately indicate pulses--don't ask me, of all people.
It has one clever idea for detection of occasional pulses. A three-position switch selects: only "1's" presented, only "0's" presented, or both are presented. Therefore, if you are looking at a line which is normally high for long periods of time, you can turn the "1" off and wait for the beep. The Radio Shack probe will do the same in the "pulse" mode, but you have to know when to put the switch there. This one, in presenting a "blended tone" in normal use, will clue you in that pulses are present.
They say it switches from TTL to CMOS logic automatically. I suppose they mean that, when the logic supply is higher than 5V (and they don't say how much), the rules of CMOS prevail. Then again, maybe you're supposed to whisper what you're working on through those little holes near the back end?
Power Requirements: 5V to 15V with reverse polarity protection. Current drain is not specified.
Input Characteristics: Input impedance is 60K. Maximum input frequency is 1.5mHz. Minimum detectable pulse is 200 nanoseconds. A transition in either direction triggers a 40-millisecond pulse stretcher. The input overload protection is good to plus/minus 50V of continuous DC, and against 150V AC up to 20 seconds.
DTL/TTL Detection Levels: "0" is less than 0.9V; "1" is greater than 2.9V.
CMOS Detection Levels: "0" is 15% of VDD; "1" is 60% of VDD.
One disadvantage is that the supply cord is of the coil-cord type, in addition to being made of "tinsel wire." Using the probe stresses the wire at the clips and causes occasional breakage. I would not condemn it for this, except for the fact that tinsel wire is so beastly hard to solder.
[To solder tinsel wire, wrap its stripped end in a strand of very thin bare wire. Then, wrap this assembly in solder and tin it quickly. Make sure the terminal is generously tinned as well. Finally, put the prepared end where it is supposed to go and heat both items until the solder melts; quickness is the key element of success,
since overheating the tinsel wire will burn up its fibers and substantially weaken it.]
Of the commercial products, I don't know which I would buy; I rather like them both. If one were to look at specifications alone, the cheaper Radio Shack unit would come out far ahead. If you're afraid of bumping the switch to CMOS and getting bad results from your tests on TTL stuff, or if you're afraid you might miss the fact that there are pulses on a line, then the Production Devices probe will calm your fears. Remember, however, that these are tricksters' instruments; they are not precision test gear. If you ever really need to know the character of your signal, measure it with something more sophisticated.
* Address List
Jensen Tool Company, 1230 S. Priest Drive, Tempe, AZ 85281 Phone: (602) 968-6231.
Production Devices, Inc., 4898 Ronson Court, Suite B, San Diego, CA 92111
Phone: (619) 278-1141.
Star Micronics, Inc., 200 Park Avenue, Suite 2308, New York, NY 10017 Phone: (212) 986-6770.
by Albert Yeo
The transistor under test is plugged into a blocking oscillator circuit. Not only will the checker ascertain whether or not the transistor is OK, but whether it is PNP or NPN can also be determined. In addition, a qualitative judgment of its gain can be made; as a bias control is turned, the "gain" information is conveyed by the rate of change in the oscillator's frequency.
I still use transistors in building auditory aids. Transistors are comparatively cheap and easy to obtain. Furthermore, they are sometimes more versatile than chips; chips are usually designed for specific purposes, so that we have to use some ingenuity in adapting them to our requirements.
This article describes a device which I have found very handy for many years past. The instrument will give you a lot of information about the transistor being tested:
Is it PNP or NPN? Which of its three (or occasionally four) leads are which? What is its gain (beta)?
The instrument uses a basic good-old "Audicator" (blocking oscillator) circuit into which the transistor under examination is plugged. You can use a standard transistor socket--or, as I do-- make your own. All the parts are available at Radio Shack, or any of your favorite stores.
The bias on the oscillator is adjustable with a rheostat on the base of the transistor. The extent to which this control has to be advanced in order to start oscillations tells you something about the gain of the transistor. You can measure the exact beta with a different test setup, but a rough feeling for the current gain is often all you need. If oscillations commence with a small advancement of the rheostat, current gain of the transistor is very high.
A polarity-reversing (DPDT) toggle switch is used to reverse the supply voltage. This switch enables you to know whether you have a PNP or NPN transistor.
To determine which terminals of the unknown are which, you could build into the checker necessary switching to get all the possible combinations, but this switching would be very complicated and would take up space on the front panel. I prefer to juggle the leads of the transistor around in the socket. Then again, a transistor can be damaged if one of its junctions is forward-biased across the terminals which are intended for "collector" and "emitter." Therefore, it might be wise to determine the lead configuration with a continuity tester first.
To a continuity tester, a bipolar transistor looks like two diodes in series-connected "back-to-back." The base can be viewed as the interconnection of these diodes. By trying different hookups with the tester, the base can be isolated; you know you have it with one tester lead when connection to either of the other elements with the "free continuity-tester lead" shows as a forward-biased diode. When this condition is met, the polarity of the "fixed lead" (the one not being shifted from one element to the other) matches the middle letter of the designation. If, for example, both diodes can be forward-biased with the positive continuity tester lead on the base, the transistor is N-P-N.
You can very often tell a power transistor by its size and/or by its shape. However, there are units in the standard TO5 can that are capable of handling 0.5 amps or so; no physical clue to this may be apparent. However, power transistors generally do not have very high current gains. Therefore, if you have to advance the rheostat almost fully to obtain oscillations, or if oscillations occur over only a small portion of the rheostat, it is probably a power transistor.
Finally, the transistors in your junkbox may have parametric "fingerprints." If in collecting transistors, for example, you have gotten some which you know to be high-frequency, and others which you know to be high-power switching units--both of which kinds are physically identical--you can very often recognize which are which by trying them in the tester.
The ways to extract information about the unknown from its behavior in this circuit cannot be easily taught; it is a matter of deduction, plus experience that you will only gain by "playing" (and maybe doing a little damage). This is essentially a "first-strike" tester; another article could deal with circuits for the accurate measurement of various parameters.
(In order to make this diagram a little easier to picture, the oscillator is described around an NPN transistor; the polarity-reversing switch is taken up later.) The collector goes through 22 ohms to the 9V line (normally positive). The emitter goes to the center-tap of a transformer primary. The bottom of this primary winding goes to the opposite 9V line (normally negative).
The top of the primary goes through a capacitor to the base (from 0.01 to 0.1 of). Shunted across the primary winding, from one end to the other, is 0.1uF. The transistor base also goes through 1K, then through RV1 (a 1megohm rheostat) to the normally positive 9V line (the 9V line closes to the collector).
The transformer secondary goes to the voice coil of a speaker which matches its impedance (usually 8 ohms). If you put about 0.1uF across the speaker, the tone will be more mellow, but I prefer the sharper sound. Changing the capacitor on the base of the transistor will change the ranges of pitches over which the oscillator works; decreasing this capacitor (to perhaps 0.01) makes the oscillator work to a high pitch.
A double-pole double-throw toggle switch is used to select the NPN ("normal") or PNP modes. If this switch is of the "on-off-on" type (with a center-off position), no on-off switch need be provided. The positive side of the 9V battery goes to the swinger of pole A; the negative side of the battery goes to the swinger of pole B. Ignoring the possible center-off position, position 1 of A goes to position 2 of B; position 1 of B goes to position 2 of A. The normally positive 9V line of the instrument goes to position 1 of A, while the normally negative line goes to position 1 of B. Where a "center-off" position is not available, an additional SPST switch (perhaps on the back of the rheostat) can be inserted in the positive battery lead.
Assuming you know, or have ascertained, which of the transistor's leads are which, plug it into the instrument; throw the switch to apply power. Adjust RV1 for oscillations. If no signal is heard, throw the switch to the other polarity and again adjust RV1. You should be able to obtain a signal with the polarity switch in one of its positions. If you cannot, you have a duff transistor.
You can mark the polarity switch as to which position is which. Start out with a transistor of known polarity; when you get it to oscillate, mark that position for this polarity, and mark the other position for the opposite transistor type.
- 1--0.1uF, disc or mylar
- 1--0.01uF to 0.1uF, disc or Mylar
- 1--1K 1/4 watt, 5% resistor
- 1--1megohm rheostat (RV1)
- 1--transistor output transformer, 500 or 1K ohm center-tapped primary, 8 ohm secondary (Radio Shack 273-1380)
- 1--DPDT toggle; if possible, with a center-off position
- 1--SPST switch (if the above has only two positions); can be included on the back of RV1.
- 1--small 9V battery
- 1--good-quality standard transistor socket
by Tom Fowle, WA6IVG
Use of different braille slates, which makes construction easier, is discussed. In addition, a technique for determining the electrical rotation and linearity of unknown potentiometers is described.
Alternative Braille Slates
Using aluminum or steel slates presents the builder with a significant problem--how can these be securely fastened to the sliding block? A solution to this problem is to use a plastic slate, made by APH. This slate has a bottom, or die, side which is thick enough to permit countersinking for flat-head screws. These slates may be screwed, rather than cemented, to the block. I have tried both self-tapping and machine screws, and both work. This slate is made of plastic which is probably easily broken, so care should be taken to avoid bashing it when it is mounted and hanging out over the edge of the machine.
Mount the slate at the end of the block closest to the pot so that you can make the smallest possible dial.
[Albert Yeo sent in an interesting idea for eliminating the need for the slide mechanism to which the slate is mounted. He suggested that multi-line slates are available which, when mounted solidly above the pot shaft, would permit dials to be made of various sizes. Available diameters would be in increments of about 16 millimeters, but this might be just fine for your purposes. He recalled that RNIB used to make 8-line slates; check around your various suppliers and see what you can come up with.]
Measuring Rotation and Linearity of Unknown Pots
After completing the article on my braille dial maker (SKTF Winter 1984), I devised a good way of determining the angle of rotation of unknown potentiometers. (The original article said that this could be done with a protractor and a continuity tester--and left it at that. If you don't know how to do it this way, don't tell anybody, since I have found a better solution.) Furthermore, my new scheme allows you to verify the linearity of a pot which you intend to use for dialage.
For pots which do not come with a nice set of detailed specifications, you must determine the angle of rotation and set this into the dial maker so that the device will make a dial having the correct overall arc length. Of course, a pot may be simply installed in the dial maker instead of the one-turn pot which normally carries the blank; the machine will make a dial for the exact component. However, this is an inconvenience, and where the electrical rotation of the pot is known (and its linearity is reliable), it is certainly easier to use the one-turn pot already in the machine.
I made a large dial which covers the full 360dgs of a circle and which has a diameter of about 4 inches. This dial contains 100 divisions. When affixed to a panel, or other firm backing, and equipped with a 3/8-inch center hole, most pots which you want to use can be mounted temporarily on this dial. You will need to make a knob and pointer for the dial; if possible, replace the set screw in the knob with a thumb screw for ease of adjustment.
Now, with a continuity tester connected between the arm and one end of your pot, turn the shaft-all the way counterclockwise to the mechanical stop, and then advance to the point where the tester just begins to change pitch. Put the pot just before this point--this being the beginning of the pot's electrical rotation. Carefully fit the knob onto the shaft so that the pointer is at the 0 mark on your dial; carefully tighten the set screw. Check to be sure you haven't moved the pot, and that 0 is really electrical 0.
Crank the pot around clockwise and find the point, just shy of the clockwise mechanical stop, where the pitch change in the tester has just ceased. This is the end of electrical rotation. The dial can now be read as "percentage of full circle." This number can be plugged directly into the dial maker for setting the "top of scale adjustment."
If you have a pot of which you are particularly fond (stranger things have been known, I suppose), and you are not sure of its linearity, you can mount it on this reference dial. With a little arithmetic, and a good ohm meter, you can now determine these facts before trying to make a dial for it. In some cases, doing all this may be more trouble than simply putting the pot in the dial maker and building a dial which fits it. However, it is always nice to have more than two ways to approach any problem; one of them may be of some real use.
For pot checking, as just described, the resistance bridge (SKTF Spring 1984) is ideal. If you haven't yet built one, get busy.
[I guess if I put this in the Summer issue, you'll have it by Christmas.]
An excellent gift idea, this simple circuit produces a realistic sound of a steam train's whistle. It's just the thing to augment Junior's train set (or Senior's, for that matter).
Back in the days when toy trains were big and heavy enough to mar the furniture and to break Daddy's ankles, we used to be able to buy electric whistles; they consisted of a squirrel-cage blower and a couple of flutes. Like everything else connected with those train sets, they were almost realistic--at least there was little doubt as to what they were intended to represent. While playing around with David Plumlee's organ chip (see "Singing Chips," (SKTF Winter 1982), I stumbled across the winning combination that puts those old relics to shame.
The circuit, although containing three chips, is quite simple; a 555 timer chip "clocks" the MK50240N organ chip (available from Jameco for about $7), with the desired notes being fed through decoupling resistors to the input of an LM386 audio amplifier. Two 9-volt batteries are used because the voltages required for the chips are so fussy (the organ chip needs 11 to 16 volts", while the LM386 cannot handle over 12 volts). Anyhow, the supply to the organ chip (and its clock) is put through a resistor and is bypassed so as to achieve the necessary "portamento" and decay characteristics needed to simulate the varying pitch of the steam whistle. (It would not do to have the 386 supply voltage change in accordance with "sagging" of the supply for the other chips.)
[I really went crazy in designing this thing at first. I had a pitch-control pot controlling the charge current to the 555 oscillator, and I also had this pot feeding LED current to an H11F3 opto-isolator whose photo FET was in an audio pad network to vary the volume as "The Conductor" played the whistle with style. The complexity of the project soon got out of hand (I burned out my only H11F3 in my zeal), and the straightforward approach again-seemed attractive. Another extreme would be to design a PLL detector that could control the whistle (which might be mounted in a boxcar, for example) via a high-frequency control signal superimposed on the power to the tracks. One could keep going forever.]
The clock circuit shown produces a signal of about 250kHz, which causes the organ chip's outputs to land in the fifth octave (more or less beginning with fifth-octave C). The notes chosen were C (pin 16), E flat (pin 6), G (pin 10), and B natural (pin 14). (Pin connections and other pertinent data are given at the end of this article.)
The unit can be mounted in a box (perhaps 1-1/2 by 2-1/2 by 5 inches), or it can be built into a piece of train paraphernalia. More important information is that the circuit will easily fit on a piece of perforated Vector board which is 1-1/4 inches wide and perhaps 4-1/2 inches long. I have my chips running lengthwise along the board; as viewed from the component side, pins 1, 2, 3 ... of all the chips are nearest the edge which carries the ground bus.
A pushbutton that controls the whistle can either be mounted on the unit, or it can be connected via a long cable. At present, I am stuck with a double-pole switch (and hence, a four-wire cable) because of the two battery supplies. There are ways of getting around this, not-the least of which is using a VMOS power FET as the "second switch" (the ambitious designer can find sufficient info in "Save Your Batteries With a Timer Switch," SKTF, Spring 1983).
Two 9V batteries are connected in series; the negative end of this stack is grounded. The positive side of the first battery goes through a pole of the normally open DPST pushbutton switch to the 9V line (provided for the LM386). This 9V line is bypassed by 1000uF (negative of the capacitor at ground). The positive side of the second battery goes through the other pole on the switch, then through 220 ohms to the "saggy" 16V line. This 16V line is bypassed by 250uF (negative at ground).
Pin 1 of the 555 timer chip is grounded; pins 4 and 8 are tied together and go to the 16V line. Between pins 1 and 8 (located close to the chip) is 0.1uF (disc ceramic). Pins 2 and 6 are tied together and go through 220pF (mica) to ground. Pin 6 also goes through 5.6K to pin 7; pin 7 goes through 6.2K to the 16V line.
Pin 3 of the 555, its output, goes to the clock terminal of the organ chip, pin 2 of the MK50240N. Pin 3 of the 50240 is grounded, while pin 1 goes to the 16V line.
Pins 2 and 4 of the LM386 are grounded, while pin 6 goes to the 9V line. Pin 7 is bypassed by 1uF (negative at ground). Between pins 4 and 5 is 0.1uF. Pin 5 goes to the positive side of a 47uF capacitor, with the negative side of this unit going through the speaker to ground.
Pin 3 of the 386 goes through 0.05uF (disc or mylar), then through 4.7K to ground. The j unction of this capacitor and resistor also goes to one end of four 560K resistors; the far end of one 560K unit goes to pin 16 of the 50240, the far end of another goes to pin 6, another to pin 10, and the far end of the fourth goes to pin 14. This completes the circuit.
For an afternoon's fun, I recommend that everyone at least breadboard this arrangement. To keep peace in the family (which will be unlikely), change the 4.7K resistor on the input of the 386 to a 10K volume control. (The bottom of the control will be grounded, the top will go to the adjoining 560K resistors, and the arm will go through the 0.05uF cap to pin 3 of the 386.) Then, put a rheostat of perhaps 10K (a pitch control) in series with the 6.2K charging resistor on the 555 clock. Finally, arrange it so that you can plug as many as five or six 560K resistors into outputs of the organ chip. Who knows, you might discover "The Lost Chord" (if the police don't discover you first). Of course, I would never advocate piping this arrangement through your 100 watt hi-fi system--no, not me. (But when you do, leave that 4.7K resistor in place, since the raw outputs from the organ chip have a voltage swing of 16V.)
- 1--220pF mica
- 1--0.05uF disc or mylar
- 2--0.1uF disc
- 1--1uF 10V electrolytic
- 1--47uF 10V electrolytic
- 1--250uF 16V electrolytic
- 1--1000uF 10V electrolytic
Resistors (1/4 watt, 5%):
- 1--220 ohm
- 1--6.2K 4--560K
- 1--LM386 audio amplifier
- 1--NE555 timer (clock)
- 1--MK50240N top-octave generator, available from Jameco*
- 1--speaker, 2 inches or larger
- 2--9V batteries
Pin Connections for the MK50240N Organ Chip
(Note: This chip requires a supply voltage of 11 to 16 volts, and it is darned fussy about it. It will seem to work, for a minute, on a 9V battery. As soon as it warms up, it'll usually quit.)
- Pin 1--plus V
- Pin 3--ground
- Pin 2--clock input
- Pin 16--1ow C
- Pin 4--C sharp
- Pin 5--D
- Pin6--D sharp (E flat)
- Pin 7--E
- Pin 8--F
- Pin 9--F sharp
- Pin 10--G
- Pin 11--G sharp
- Pin 12--A
- Pin 13--A sharp
- Pin 14--B
- Pin 15--high C
* Jameco: 1355 Shoreway Rd., Belmont, CA 94002; Phone: (415) 592-8097.