TABLE OF CONTENTS

NOTES ON BLOOD-PRESSURE MEASUREMENT

SMITH-KETTLEWELL BLOOD-PRESSURE GAUGING CIRCUITS

A NOVEL VCO CIRCUIT USING A GARDEN-VARIETY OP-AMP

BULLETIN BOARD

EDITOR'S CORNER

NOTES ON BLOOD-PRESSURE MEASUREMENT

[Acknowledgments: Since the editor knew nothing about this subject, much credit is due the following experts: Eileen Abbott, a physical therapist at Monroe Community Hospital in Rochester, New York; Ms. Bonnie Jo Bell, Coordinator at the Cardiac Rehabilitation Unit, Pacific Presbyterian Medical Center, San Francisco; Dr. T. A. Benham of Science Products in Berwyn, Pennsylvania; Deborah Churchill, Director of the Visually Impaired Training Program at St. Mary's Junior College in Minneapolis, Minnesota; and Dr. Alvin H. Sacks at the VA Medical Center in Palo Alto (Dr. Sacks having done a great deal of work on "indirect blood-pressure measurement" in his earlier careers). (Thanks Ms. Bell; it's a lot easier after you taught me to use my arm ... instead of my neck?)]

The real way to measure blood pressure is to tap into an artery. If this idea makes you a little faint, don't worry, there are ways of getting blood-pressure information indirectly--without invading the flesh. The most traditional is by pressurizing a tourniquet around a limb until an artery is mostly collapsed, then making judgments about blood flow as the tourniquet is gradually relaxed. A gauge attached to the inflated tourniquet allows you to read air pressure--pressure inflating the "cuff," not pressure of the pulsing blood.

Around the turn of this century, a Russian named Korotkoff formalized the technique of listening to sounds of the blood vessel with a stethoscope while relaxing the pressure in the tourniquet (cuff). Enough air is pumped into the cuff to collapse the artery (although it is never completely closed off). As the pressure is slowly decreased, the point at which peaks in blood pressure cause the artery to open sufficiently to make noise is noted, and the air gauge is read at that point. Further relaxation of the cuff brings about noisy instability in the walls of the artery. Eventually, these sounds give way to a duller pulsing rumble; from this point on, the artery walls never touch each other, and a cross-sectional view looks like an ellipse whose minor axis elongates with pulses of blood flow. Finally, the sounds become more or less inaudible as the artery approaches its normal cross-section.

Opinions differ as to when the second reading is taken. Dr. Sacks tells me that the British consider that point to be when "Korotkoff sounds" suddenly fade, while the U.S. standard calls for the sounds to disappear.

The first reading is called the "systolic" pressure; the blood pressure is at its maximum, caused by ventricular contraction. The second reading is called the "diastolic" pressure, occurring when the heart is filling with blood.

The Korotkoff sounds are only one way of doing it. Dr. Sacks recommends, in taking your own blood pressure, that you attend to palpitations (throbbing) of the vessel in the arm. The systolic reading is taken when bursts of blood flow first make the vessel's instability noticeable. A phase then passes during which you feel a kind of "internal tapping" as the vessel walls open and collapse. Further deflation allows the palpating to subside, and practice with your doctor will help you decide when the diastolic reading should be taken.

Many electronic instruments use a microphone in place of the stethoscope; they try to replicate a person taking blood pressure using the Korotkoff sounds. Some of the newer electronic blood-pressure meters get their information from pressure waves in the cuff. These waves, caused by pulsing under the cuff, are different from Korotkoff sounds, which are monitored just below the cuff. This constitutes yet another kind of analysis.

There seems to be some controversy over differences in these techniques of indirect measurement. Each, in and of itself, may be consistent, but there is an undercurrent of doubt as to their equivalency.

The gauge circuits in the succeeding article are of the simplest kind; these gauges are not equipped to analyze sounds or pressure waves. They are made for users who can either interpret the Korotkoff sounds, or for those who want to measure their own blood pressure by interpreting vessel palpitations (rather than sounds). Accompanying each of our pressure-gauge boxes (which are about 3-1/2 by 6 by 2 inches), a standard inflatable cuff is used. If Korotkoff sounds are to be heard, a stethoscope is also required; Cardiac Rehab Coordinator Bonnie Jo Bell tells me that the proper set-up employs a stethoscope which is separate from the cuff.

Basic Procedure Using Korotkoff Sounds

Either arm can be used. The sleeve should be rolled up; if this cannot be done without constricting the arm, clothing should be removed so as to eliminate constriction and expose bare arm up past the elbow. The cuff is loosely wrapped around the upper arm, with its lower edge positioned 1 inch above the "antecubital space" (the hollow inside the elbow).

The stethoscope is placed over the brachial artery (which runs through the hollow of the elbow). (Some stethoscopes can be switched from a diaphragm to an open cup; what you want to have is the diaphragm working into a closed air column.) To find out just where to place it, you can locate your own brachial artery by feeling for its pulse with a couple of fingers in the crook of the elbow, along the edge of the hollow nearest your body.

Finally, the arm should be comfortably supported, more or less level so that it is level with the heart.

Using the gauge to monitor inflation pressure, the cuff is pressurized to the extent that assures cutoff of the major blood flow--200 millimeters of mercury might be enough, but 250mm or more will be required in cases of hypertension. The author selects Position 4 of the function switch on our Smith-Kettlewell gauges, makes sure that the audible readout is active, and sets the Braille dial for 200mm; then the cuff is inflated (to the accompaniment of a rising tone) until the tone begins to pulsate, signifying that the dial's setting has been reached.

Next, the "bleed" (an adjustable leak, usually a button or thumb-screw valve at the bulb) is adjusted for a slow descent in cuff pressure (the rate of which can be heard as a descending pitch in our gauges). Careful listening to the stethoscope will alert the user to the initial Korotkoff sound--the first pulse of blood that can be heard. (This must be distinguished from squeaking of the elbow and other extraneous noises.)

The systolic pressure is noted as being the value at which the first of those special sounds (like a faint thumping--one per heart beat) is heard. On our gauges, a button is pressed to sample and store that value of pressure. Those sounds will subsequently become prominent. The diastolic reading is to be taken after their character has changed to a very dull throb, or when they virtually fade away, depending on whose definition you accept. A second button on our gauge is to be pressed to sample that pressure.

The rate of deflation, while not absolutely critical, can, in the extreme, affect the readings. In our comparative experiments with Ms. Bell, it was noted that a fast deflation rate led to low readings; pressing of the button was more often late as sounds were verified. On the other hand, wearing the tourniquet affects blood flow and artificially raises the blood pressure, so the deflation can be too slow. We were getting fairly good results at a deflation of 2.5mm per second.

One notable difference between blood pressure readings taken by the author and by Ms. Bell was that, since she could remember where the visual gauge was when she suspected initial Korotkoff sounds, she could, in effect, think back to note the reading after it had been verified by subsequent sounds. I, on the other hand, really committed myself when I pushed the button--usually done at the point of verification. Thus, my readings tended to be lower than hers. Of course, this difference was markedly larger when the deflation rate was rapid (8mm per second).

Although I haven't decided how to accomplish it, a gauge should be built with a little synchronous memory--synchronized to the pulse rate found in pressure waves in the air hose--which would allow you to press the button on the third Korotkoff pulse, and then direct the sample to be at the first. A nice idea, but probably hard to implement.

Indirectness of Indirect Blood-Pressure Measurement

An illuminating paper is one by Alvin Sacks, Ph.D., called "Indirect Blood Pressure Measurements: A Matter of Interpretation" (Angiology, Vol. 30, No. 10, October 1979). In that paper, a simulation of auscultatory blood pressure measurement, using 1-inch rubber tubing and a pulsating water pump, is documented. A number of variables that obscure the relationship between indirect measurement and actual system pressures are listed: For starters, what you are listening to is the dynamic instability of the brachial artery, and this is profoundly affected by the artery's radius, its wall thickness, the elastic modulus of the arterial wall, and its "axial tension in vivo" (which is my favorite). The cuff length even makes a difference; a narrow cuff gives you higher readings than a wide one.

Two of the above variables markedly change with age: the elastic modulus decreases, and the wall thickness increases. A main thrust of Dr. Sacks' paper was to demonstrate the profound effect that wall thickness has on auscultatory readings. A doubling of the thickness, apparently not unusual, can cause an error of 30mm (to the high side), and there is such a thing as "pseudo-hypertension" in older patients, caused by this thickening.

Well then, what are we doing here; if we're not likely getting true readings of blood pressure, let's have a salami and forget about it. Actually, there is a point to it, based on a reservoir of data over a period of 80 years. Even if we dismiss these numbers as poor indications of actual blood pressure, experience with them as relative indicators will tell you if you are at risk of having a stroke, or if you need a transfusion. By now, expected readings for all ages have been tabulated, and whether they represent actual pressures or not, massive correlation of numbers to illness or "wellness" exists which gives your doctor information of historic significance.

SMITH-KETTLEWELL BLOOD-PRESSURE GAUGING CIRCUITS

Abstract

Two versions of sphygmomanometers are described: One is a low-cost instrument that we can all build for the benefit of our health and well being. The other is a lab-quality gauge with an analog output from which strip-chart recordings can be made; this output can also be used to feed a talking meter.

[Correction: In the article "A Survey of Sensym Pressure Transducers," SKTF, Summer 1988, the sensitivity and offset figures for the BP01 are given with a mistakenly implied supply of 12 volts. (Everything else in the list of transducers was "spec'ed" at 12 volts.) On the BP01, the figures given are for a 5-volt supply; no 5/12ths correction factor is needed for our "lab" gauge circuit, since we used 5 volts.]

Note: Commercial equipment is available. Even though no self-contained talking sphygmomanometer is in current production, a limited supply of instruments still exists at AFB; the price is about $125. Science Products sells a fancy computerized system. Called the "Health Kit," this talking computer console can manage four instruments: a thermometer, a sphygmomanometer, a pulse-rate meter, and a weight scale. That set of instruments is in the $800 class.

Descriptive Introduction

The simplicity of this project suggested itself in the Sensym gauge catalog; they list a transducer specifically for indirect blood pressure measurement, the "BP01." Messing around with it, I decided that any precision I was likely to achieve in operating a blood-pressure cuff was far less than that offered by the BP01, which is good to within 0.2% (at a cost of $38). Thus, why not use a $13 sensor? Two units were built having features as follows:

In both, a standard cuff arrangement, with stethoscope, is used (robbed from a kit whose gauge was bad). (New kits, with gauge, can be gotten for about $20 at discount drug stores; the gauge can be discarded, or attached with a "T" fitting for simultaneous visual reading.)

Using the Sensym SX05DN sensor, the low-cost gauge was built into a 3-1/2 by 6-1/4 by 2 inch cabinet. Nipples from the sensor protrude through the top at the left front corner; the hose from the cuff is plugged onto the nipple of Port 1, the port nearest the printed face of the SX05DN.

On both instruments, all of the controls and switches are mounted on the top panel. Two sample-and-hold pushbuttons are used to "note" the readings. A 4-position rotary switch, which also takes care of the on-off function, switches a standard auditory meter circuit between "Sample 1," "Sample 2," or the gauge directly.

Mr. Fowle's fine meter circuit (termed the "Fowle Gimmique" in other writings) is used to take the measurements. A pointer knob and Braille scale take up the right-hand 2/3 of the panel. Finally, a silence switch in series with the speaker allows you to cut off the sound of the meter reader, if desired; this switch is in the upper-right-hand corner of the top panel.

The range of cuff pressures measured by the low-cost unit was chosen to be from 50 to 250 millimeters of mercury. Single dots mark every 10mm, double dots mark 50, 150 and 250mm, and triple dots are placed at 100 and 200mm.

Physically, the "laboratory" instrument looks almost the same as its junior partner. An insignificant difference is that the shape of the BP01 sensor lends itself to gaining access to the nipples from the front panel (instead of the top); Port A is used, which is the one nearest the BP01's pins. The controls on the top are in the same places; the 4-position rotary switch and the "sampling" buttons are at the left end, and the Braille dial takes up the right 2/3 of the top panel. A BNC connector permitting access to the analog gauge signal is mounted on the rear panel; this was provided so that strip-chart recordings of measurements could be made, or so a talking meter could be attached.

I was advised by physical therapists that blood-pressure readings on hemorrhaging patients require that the gauge work below 50mm of mercury (poor devils). AMA standards also insist that the upper end of the range be 300mm.

On the Braille dial of the lab model, single dots mark every 10mm, double dots are placed at 50, 150 and 250mm, and triple dots mark the hundreds. If I had it to do over again, I would have built this model in a larger cabinet, since a Braille scale of this density should have a radius of perhaps 2-1/2 inches.

Note that, as far as the VCO is concerned, you will get an output for pressures outside the ranges of the instruments--probably to as much as 500mm of mercury. For your own use, then, as long as you're alive and able to run the equipment, I see no disadvantage in the cheap gauge.

Two 9V batteries are used to create a dual supply. (While this is a nuisance, the negative supply handles so little current that when the positive-supply battery gets weak, just trading their positions will make the gauge operable again.)

Operation of the Gauge

(Taking blood pressure is one matter; running the gauge is another. For more on accomplishing indirect blood-pressure measurement, see the previous article.)

Both instruments are operated in exactly the same way. The audible output is a VCO whose pitch rises as a function of cuff pressure. Readings are taken by setting the Brailled pointer knob to the point, or "boundary," where the tone becomes smooth and continuous on the high side, versus being "pulsed" on the low side. In other words, the tone either pulsates or is continuous; the point of change-over is the control position where the Braille scale is read.

While "taking the blood pressure," it is a matter of choice as to whether the readout is silent or operative; the silence switch cuts off the loudspeaker. (In any of the three "on" positions, the sample-and-hold buttons work, and will record your readings.) On the other hand, with the selector switch in position 4 and the readout made operable, you can set a target for maximum inflation and monitor the bleed-off rate by listening to the VCO. The author sets the pointer to 200mm, and with the meter reader operative, inflates the cuff until the VCO's tone pulsates. I start bleeding the cuff, noting the descending pitch as a measure of the rate of deflation. Then, I silence the readout so as to listen to the stethoscope. (Our experiments show that a slow rate of deflation leads to more accurate readings than a fast rate, since any delay in pressing the buttons will make the readings low. On the other hand, too slow a deflation rate will, by changing circulation of blood, artificially raise the blood pressure readings. About 2-1/2mm deflation per second--4 seconds per 10mm division--is optimal.)

As air is bled out of the cuff, you set yourself up to press the pushbuttons--one when you first hear pulses of blood flow (the systolic reading), and the other when you are satisfied that the diastolic reading should be noted. These buttons "store" your readings in sample-and-hold capacitors. When you are through with the cuff, the two readings are taken with the rotary switch in positions 2 and 3. (It doesn't matter which button--and thus which corresponding switch position--you use for systolic and diastolic sampling; the circuits are identical.

Important! When taking your own blood pressure, use the other hand--the one not involved with the arm cuff--to push the buttons. If you forget and use the same arm to operate the instrument, you'll think the Grim Reaper is about to trip your circuit breaker and perform a "cold boot."

Circuits

(Note: Tom Fowle figured out a way of eliminating the current mirror--the two PNP transistors in his readout, the Fowle Gimmique. By George, it's clever, and I suggest that you properly attend to it.)

Important for the Sample-and-Hold! The rotary switch mentioned in these circuits must be of the "break-before-make" type. My Radio Shack 3-pole 4-position switch turned out to be of the "shorting" type ("make-before-break"), and every time I switched to a sampling position, I destroyed the sample as adjacent positions briefly encountered one another. You can tell if a switch is of the wrong kind with a continuity tester; with the tester connected to adjacent positions, a brief beep as the switch is slowly turned between these positions will tell you that you have the undesired kind.

The choice of an op-amp with FET inputs is absolutely critical in the readout circuit. If you try an LM358 in this service, the input currents will cause the samples to drift too rapidly to read. An RCA CA3240 has input currents in the range of nanoamps, and the sampled voltage is only very slowly degraded by its invasion.

Low-Cost Gauge Circuit

A Sensym SX05DN 5psig gauge is used, working into an LM324 quad op-amp. The gauge runs off a dual 9-volt supply. The negative side of one 9V battery is grounded; its positive terminal goes to positions 2, 3 and 4 of deck "A" on a 3-pole 4-position rotary switch. The positive terminal of another 9V battery is grounded; its negative terminal goes to positions 2, 3 and 4 of deck B. The arm of deck A goes to the plus 9V line of the gauge circuit. The arm of pole B goes to the project's minus 9V line.

Pin 4 of the LM324 goes to the plus 9V line; pin 11 goes to the minus 9V line. An LM336 is used to create a 2.5V line. The anode of the 336 is grounded. The cathode goes through 2.2K to the plus 9V line; this cathode is the 2.5V point.

Pin 5 of the LM324, a non-inverting input, goes to the 2.5V point. Pin 6, the inverting input, goes through a 10K 1% resistor to ground. Between the 324's pins 6 and 7 is a 10K 1% feedback resistor. This pin 7 becomes a 5V supply point; it goes to pin 3 of the SX05 sensor. Pin 1 of the sensor is grounded.

Another op-amp in the 324 makes an adjustable supply. Pin 3, the non-inverting input, goes to the arm of a 50K trim pot. The top of this pot goes to 2.5V; the bottom of the pot is grounded. Pin 2, the inverting input, goes through a 100K 1% resistor to ground. Between pins 1 and 2 is a 10K 1% feedback resistor.

The remaining two op-amps in the LM324 comprise the differential amplifier for the sensor. Pin 13, an inverting input, goes through a 10K 1% resistor to pin 1, the output of the above adjustable supply. Between pins 13 and 14 is a 10K 1% feedback resistor. The output of this op-amp, pin 14, goes through another 10K 1% resistor to pin 9, the inverting input of the last amp in the package. Between pins 8 and 9 is a 10K 1% feedback resistor.

The gain is determined by a 4.02 or 4.32K 1% resistor between the inverting inputs--between pins 9 and 13. (The actual value of this resistor is not critical; it should be of good quality, and hence the specified tolerance of 1%. Our resistor was made up of 1K in series with the parallel combination of three 10K 1% units.

The sensor's pin 2 goes to the positive input of the differential amplifier--pin 10 of the 324. Pin 4 of the sensor goes to pin 12 of the 324. The output of the gauging circuit is the 324's pin 8. This output gives you 1.98 millivolts per millimeter of mercury--496mV for 250mm full scale.

This pin 8 goes to position 4 of deck "C" on the selector switch. Pin 8 also goes to one side of two normally open SPST pushbuttons. The free terminal of each pushbutton then goes through its own 0.47uF Mylar capacitor to ground. (In other words, pressing a button connects pin 8 to a 0.47uF sampling capacitor.) The top of one capacitor--its junction with the associated pushbutton--goes to position 3 of deck C. The top of the other capacitor goes to position 2.

The arm of deck C goes to the readout circuit which follows:

The metering section uses an RCA CA3240 dual op-amp; one op-amp in the package is part of a voltage-controlled current source for the VCO, and the other is a comparator for the Brailled pot. Pin 4 of this 3240 is grounded. Pin 8 goes to plus 9V. The non-inverting inputs, pins 3 and 5, are tied together and go through 220K to the arm of switch deck "C."

Pin 6 of the 3240 goes to the arm of a 10K linear precision pot which is fitted with a pointer knob and Braille scale. The bottom end of this pot goes through a 2.5K 1% resistor to ground. (Although a 1% value of 2.49K is available, we made a 2.5K resistor out of four 10K 1% resistors in parallel.) The pot's top end goes through a 10K 1% resistor, then through a 50K calibration rheostat to the 2.5V point.

On the CA3240, the pin 1 output goes to the base of a 2N2222 NPN transistor. The emitter of this transistor goes through 2.2K, then through a 50K rheostat (VCO sensitivity) to ground. This emitter also goes to pin 2 of the 3240, the inverting input of this section.

The collector of the 2N2222 goes through 22K, then through 10K (both 5% resistors) to pins 2 and 6 of an NE556 dual timer--2 and 6 are the Threshold and Trigger pins. Pins 2 and 6 also go through 0.0047uF to ground. The junction of the 22K and 10K resistors goes to the cathode of a 1N914 diode; the anode goes to the output, pin 5.

Pin 4 of the 556, the Enable pin of the above VCO, goes to pin 9 on the same chip, the output of the pulsation oscillator section. Pins 8 and 12 of the 556 (Trigger and Threshold) are tied together and go to the positive end of a 2.2uF capacitor; the negative end of this cap is grounded. Pins 8 and 12 also go through 10K to pin 13, the Discharge pin. Pins 8 and 12 also go through 22K to the output of the comparator, pin 7 of the CA3240.

Pin 7 of the 556 is grounded; pin 14 goes through 10 ohms to the plus 9V line. Pin 14 is bypassed to ground by 100uF (negative end of this cap at ground). The Enable of the second section, pin 10, goes to pin 14.

The output of the oscillator, pin 5, goes to the positive end of a 10uF capacitor; the negative end goes through a 100-ohm resistor to one side of the SPST "silence" switch. The other side of this switch goes through the speaker to ground.

Lab-Quality Gauge Circuit

A Sensym BP01 sensor is used, working into an LM324 quad op-amp. The gauge runs off a dual 9-volt supply. The negative side of one 9V battery is grounded; its positive terminal goes to positions 2, 3 and 4 of deck "A" on a 3-pole 4-position rotary switch. The positive terminal of another 9V battery is grounded; its negative terminal goes to positions 2, 3 and 4 of deck B. The arm of deck A goes to the plus 9V line of the gauge circuit, and this is bypassed to ground by 100uF (negative of this cap at ground). The arm of pole B goes to the project's minus 9V line, and this is bypassed to ground by 10uF (positive of this cap at ground).

Pin 4 of the LM324 goes to the plus 9V line; pin 11 goes to the minus 9V line. An LM336 is used to create a 2.5V line. The anode of the 336 is grounded. The cathode goes through 2.2K to the plus 9V line; this cathode is the 2.5V point.

Pin 5 of the LM324, a non-inverting input, goes to the 2.5V point. Pin 6, the inverting input, goes through a 10K 1% resistor to ground. Between the 324's pins 6 and 7 is a 10K 1% feedback resistor. This pin 7 becomes a 5V supply point; it goes to pin 2 of the BP01. Pin 4 of the BP01 is grounded.

Another op-amp in the 324 makes an adjustable supply. Pin 3, the non-inverting input, goes to the arm of a 50K trim pot. The top of this pot goes to 2.5V; the bottom of the pot is grounded. Pin 2, the inverting input, goes through a 100K 1% resistor to ground; pin 2 goes through another 100K 1% resistor to the 2.5V point. Between pins 1 and 2 is a 10K 1% feedback resistor.

The remaining two op-amps in the LM324 comprise the differential amplifier for the sensor. Pin 13, an inverting input, goes through a 10K 1% resistor to pin 1, the output of the above adjustable supply. Between pins 13 and 14 is a 10K 1% feedback resistor. The output of this op-amp, pin 14, goes through another 10K 1% resistor to pin 9, the inverting input of the last amp in the package. Between pins 8 and 9 is a 10K 1% feedback resistor.

The gain of this differential amplifier is determined by a 500-ohm 1% resistor between inverting inputs--between pins 9 and 13. (The actual value of this resistor is not critical; it should be of good quality, and hence the specified tolerance of 1%. Ours was made of two 1K 1% units in parallel.)

The sensor's pin 5 goes to the positive input of the differential amplifier--pin 10 of the 324. Pin 3 of the sensor goes to pin 12 of the 324. The output of the gauging circuit is pin 8. This output gives you 2.1 millivolts per millimeter of mercury--630mV for 300mm full scale.

This pin 8 goes to position 4 of deck "C" on the selector switch. Pin 8 also goes to one side of two normally open SPST pushbuttons. The free terminal of each pushbutton then goes through its own 0.47uF Mylar capacitor to ground. The top of one capacitor--its junction with the associated pushbutton--goes to position 3 of deck C. The top of the other capacitor goes to position 2.

The arm of deck C of the rotary switch goes to the readout circuit which follows:

The metering section uses an RCA CA3240 dual op-amp. Pin 4 of this 3240 is grounded. Pin 8 goes to plus 9V. The non-inverting inputs, pins 3 and 5, are tied together and go through 220K to the arm of deck "C."

Pin 6 of the 3240 goes to the arm of a 10K linear precision pot which is fitted with a pointer knob and Braille scale. The bottom end of this pot is grounded; its top end goes through a 10K 1% resistor, then through a 50K calibration rheostat to the 2.5V point.

On the CA3240, the pin 1 output goes to the base of a 2N2222 NPN transistor. The emitter of this transistor goes through 2.2K, then through a 50K rheostat (VCO sensitivity) to ground. This emitter also goes to pin 2 of the 3240, the inverting input of this section.

This 3240's pin 2, which contains a "virtual version" of the gauging signal, goes to the hot side of a BNC output connector. The cold side of this connector is grounded.

The collector of the 2N2222 goes through 22K, then through 10K (both 5% resistors) to pins 2 and 6 of an NE556 dual timer--2 and 6 are the Threshold and Trigger pins. Pins 2 and 6 also go through 0.0047uF to ground. The junction of the 22K and 10K resistors goes to the cathode of a 1N914 diode; the anode goes to the output, pin 5.

Pin 4 of the 556, the Enable pin of the above VCO, goes to pin 9 on the same chip, the output of the pulsation oscillator section. Pins 8 and 12 of the 556 (Trigger and Threshold) are tied together and go to the positive end of a 2.2uF capacitor; the negative end of this cap is grounded. Pins 8 and 12 also go through 10K to pin 13, the Discharge pin. Pins 8 and 12 also go through 22K to the output of the comparator, pin 7 of the CA3240.

Pin 7 of the 556 is grounded; pin 14 goes through 10 ohms to the plus 9V line. Pin 14 is bypassed to ground by 100uF (negative end of this cap at ground). The Enable of the second section, pin 10, goes to pin 14.

The output of the oscillator, pin 5, goes to the positive end of a 10uF capacitor; the negative end goes through a 100-ohm resistor to one side of the SPST "silence" switch. The other side of this switch goes through the speaker to ground.

Note on the BNC Output

In order to make the sampled values available at this connector, a take-off point in the readout circuit was chosen instead of just the gauge output. Avoiding installation of a buffer, it was noted that a "virtual version" of the selected signal exists at the inverting input of the voltage-controlled current source (pin 2 of the CA3240). The main disadvantage in using this point is that any load present affects the VCO; a 50K load will raise the pitch by several musical tones (the degree depending on where the VCO sensitivity is set). On the other hand, though the VCO will play to the puppy dogs as a result, a load as heavy as 1K will not affect the voltage on this connector. You must consider, though, that the current drawn by your external load is directly reflected in the base circuit of the 2N2222, and this has its limits. Shorting this output may very well burn out that transistor.

Calibration

For the two models, calibration procedures are very similar. The only difference is that the range of the laboratory unit permits you to use the full scale of any visual gauge for calibration, and this will minimize the effect of error in the visual gauge. On the lab model, no provision was made to "calibrate" the BNC connector's output, since our strip-chart recorder has a vernier gain adjustment.

Turn the on-off/selector switch to position 4 and make sure the silence switch is closed. If the readout isn't singing already, adjusting the offset pot on pin 3 of the LM324 should bring it to life. With the readout singing at a comfortable pitch, blow into Port 1--or suck on Port 2--and make sure that the gauge is responding. Then, bring the offset pot down to where the VCO frequency goes to zero. (This will be a fair guess of the gross setting of this pot.)

Procure a visual sphygmomanometer, remove the hose from its gauge, and with a "T" fitting and a short piece of 3/16th rubber tubing, connect both gauges to the cuff. Install the cuff on a "dummy arm"--a deceased neighbor or a leg of your Louis XIV lab bench. (We used a 4-inch mailing tube, but the dummy could be anything; it doesn't even have to be round.)

With the "bleed" on the cuff closed, inflate the cuff to a pressure of over 250mm. The pneumatic system of the cuff will probably leak a little, causing the pressure to gradually drop. Make a rough attempt at calibration; then, bleed off the pressure until you're below 100mm and check the calibration. It is likely that good "tracking" of the readings will have to be achieved by further minor adjustment of the offset. By going back and forth between high and low pressures, iterative adjustments of the two pots will lead to a condition of good tracking. At the high value, adjust the "calibration pot"--the rheostat in series with the Braille-calibrated control. At the low readings, adjust the "offset" pot--the trimmer whose arm goes to pin 3 of the LM324.

Because these systems leak, you will probably have to do your calibration "on the fly," as you might say. You can set the Brailled pot to a value just below where the visual gauge is reading, and wait for the system to drop to that chosen level. If the VCO smoothes out (stops pulsating) as that value is crossed, your calibration is correct (at that end of the scale). Another technique is to gently squeeze the bulb so as to hold a reading; you can get pretty good at "hovering" around a reading while someone else adjusts the appropriate trimmer. Don't forget to check for good tracking of the gauges throughout the range.

Finally, the sensitivity of the VCO can be adjusted to your liking; the only requirement is that it be listenable throughout the full range. This adjustment, the rheostat in the emitter circuit of the 2N2222, has no effect on the Braille calibration of the gauge.

Board Layout

In both models, the board is 6 by 1.8 inches, and is mounted (using 4-40 3/4-inch bolts and spacers) to the front panel of the cabinet. The plus 9V bus (the main VCC bus) is the top edge--nearest the top panel. The other long edge sports the ground line. Naturally, the "wiring side" of the board faces the front panel; the "component side" of the board faces the inside of the cabinet.

The way the sensors are shaped, and thus the way they are affixed to the board, is the major difference between the two models. (Actually, you don't need to mount the sensor on the PC board; you could bolt it to the cabinet and connect it via a 4-wire cable.)

The BP01 has its nipples coming out of the mounting surface; thus, if you mount it on the board, holes must be made to accommodate these nipples, and the board will have to be within 1/8th inch of the panel so that usable portions of the nipples come through. For connection to the pins, 3/16th inch of each pin is bent at right angles in the direction of the nipples; then the pins can be plugged into a high-profile socket on the board. The sensor is at the left end of the board with the socket to its right.

The SX05DN is mounted with its nipples sticking up beyond the VCC edge of the board. For connection, 3/16th inch of each pin is bent at right angles and plugged into a low-profile socket immediately below the sensor. When the board is mounted on the front panel of the cabinet, the SX05's nipples come through the top. Here too, the sensor is at the left end of the board, leaving just enough room for one of the board's mounting bolts at the upper-left corner.

Although the circuits are different here and there, placement of the parts can be exactly the same. The LM324 is perhaps 0.6 inches to the right of the sensor. Beyond this chip, the offset and "span" trimmers are placed so that their adjustment screws face the bottom edge. To the right of these pots, the CA3240 op-amp can be put. Since this 3240 is an 8-pin ("miniDIP") package, there is room either above or below it for the LM336 regulator. To the right of the 3240, you can construct the 2N2222--up near the plus VCC bus), and the VCO sensitivity trimmer (an adjustable emitter resistor for the 2222) can be positioned with its screw facing the ground bus. Finally, the 556 logically resides at the right end of the board.

Pin Assignments

LM336 2.5V Voltage Regulator:

• With the flat side of the package toward you and with the leads pointing up, the connections are from left to right: anode (ground), cathode (plus 2.5V), and "Adjust" (not used).

BP01 Sensym Blood-Pressure Sensor:

• With the nipples facing down and the pins toward you, the pins are numbered 1 through 6, respectively, from left to right.
• Pin 1--Temp Out Plus (not used)
• Pin 2--Plus supply (to pin 7 of the LM324)
• Pin 3--Port A Minus Out
• Pin 4--Ground
• Pin 5--Port A Plus Out
• Pin 6--Temp Out Minus (not used)

SX05DN Sensym 5psi General-Purpose Pressure Sensor:

• With the printed side up and the pins facing you, the pins are numbered 1 through 4, respectively, from left to right.
• Pin 1--Ground
• Pin 2--Port 1 Plus Out
• Pin 3--Plus Supply (to pin 7 of the LM324)
• Pin 4--Port 1 Minus Out

Parts List

Low-Cost Gauge

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

• 1--10 ohm 1/2-watt
• 1--100 ohm
• 2--2.2k
• 2--10k
• 2--22k
• 1--220k

Resistors (1%):

• 1--2.49K (although this value is available, we used 2.5K, made from four 10k's in parallel)
• 1--4.32k (Although this value is available, we comprised a 4.3K assembly by putting a 1K 1% in series with the parallel combination of three 10K units)
• 14--10K (this includes seven units which were used to create odd values--see above)
• 2--100k

Capacitors:

• 2--0.47uF Mylar
• 1--0.0047uF disc or Mylar
• 1--2.2uF 10V electrolytic
• 1--10uF 10V electrolytic
• 1--100uF 10V electrolytic

Semiconductors:

• 1--SX05DN pressure sensor
• 1--LM336
• 1--LM324
• 1--CA3240
• 1--NE556
• 1--2N2222
• 1--1N914

Pots and Switches:

• 3--50k 10-turn trim pots
• 1--10K linear precision pot (Clarostat 58C1-10K) with braille dial (see text)
• 1--3-pole 4-position non-shorting rotary (Mouser 10YX034)
• 2--normally open push buttons

Miscellaneous:

• 2--9V batteries with connectors
• 1--Cabinet 3-1/2 by 6-1/4 by 2 inches
• 1--lauden schpreker (small)
• 1--Sphygmomanometer--old-fashioned with mechanical gauge and squeezing bulb.
• 1--Optional 3/16-inch "T" fitting, Player Piano Company Cat. No. 370 1--length of tubing for optional T fitting, Player Piano "expression tubing" or equivalent.
• 1--body (with or without measurable blood pressure)

Parts List

Lab Gauge

The Sensym BP01 sensor is substituted for the SX05DN. A BNC connector is added. The only other changes are in the precision resistors, which are re-listed as follows:

Resistors (1%):

• 1--500 ohms (made of two 1K units in parallel)
• 7--10K
• 2--100K

Address List

Brill Electronics (for the Clarostat pot): P.O. Box 1378, Oakland CA 94606; (415) 834-5888, ask for Pam at Ext. 26.

Mouser Electronics: 11433 Woodside Ave., Santee, CA 92071: (619) 449-2222.

Player Piano Co.: 704 E. Douglas, Wichita, KS 67202: (316) 263-3241.

Sensym: 1255 Reamwood Ave., Sunnyvale, CA 94089: (408) 744-1500.

A NOVEL VCO CIRCUIT USING A GARDEN-VARIETY OP-AMP

by Bernie Vinther

For a long time, the 555 has been used as a pretty good VCO; but it has some limitations. For one thing, without extra circuitry, it can only act on positive incoming voltages on its input. Another problem is that this DC input must exceed a threshold before the oscillator starts working. (With the CMOS version, this threshold can be lowered considerably by bringing pin 5 close to ground, but you cannot get much closer than about 150 millivolts above ground. One more problem is that the closer you bring pin 5 to ground, the greater will be the frequency-to-voltage sensitivity (which is true because you are shrinking the charge/discharge range of the RC timing circuit).

I feel that this op-amp-based VCO has some advantages over the 555 and other VCO integrated circuits. For one thing, when you have an op-amp left over in a circuit, rather than adding a VCO chip, you can make a VCO with the spare op-amp.

Some VCO chips have an open-collector output, and most cannot source the same amount of current that they can sink. The output of the op-amp is more symmetrical in its capability (although it will go into current limiting at some point--perhaps at about 7mA). Open-collector outputs require an external resistor, and this can draw a lot of current if you want a low-impedance output.

The input impedance of this circuit depends on a charging resistor, but this can be fairly large. (This impedance is dynamic; during one part of the cycle, the input resistor is used to charge a capacitor, while during the other portion of the cycle, this resistor is tied to the output through a diode.)

True, this circuit operates from a dual (or "split") supply; but, by changing the polarity of its diode, you can make this circuit respond to either a positive or negative voltage.

Circuit Operation

If you look at the internal diagram of a 555 (see SKTF, Winter 1986), you will notice that the 555 has some comparators in it whose outputs control a bistable flip-flop. This circuit uses a comparator (a 741 or any op-amp) which is arranged to have hysteresis and is therefore bistable.

Hysteresis is of key importance to making this circuit work; it causes the output of a voltage comparator to stay locked in a high or low state until the input voltage at the inverting input exceeds the voltage fed back to the non-inverting input (from a positive feedback network). Normally, only a small portion of the output voltage is fed back to the non inverting input. When the voltage at the inverting input exceeds that on the non-inverting input, the output flips to the opposite state and stays locked in this condition until the voltage at the inverting input again exceeds that which is fed back to the non-inverting input (only in the opposite polarity).

By itself, this type of positive feedback does not cause the circuit to oscillate, but only provides a stability that keeps the output locked at one saturation point or the other until the "hysteresis" is overcome by a sufficient level of input signal.

A way to look at the positive feedback network is as a voltage divider from the output to ground, with the non-inverting input looking at the take-off point. Because this provides positive feedback, the output will quickly saturate in one direction or the other; this will hold the non-inverting input at some level away from ground. To toggle the circuit--to make the output go the other way and lock in that position--the inverting input will have to be applied in such a direction and with sufficient magnitude as to exceed the magnitude that is found at the take-off point of the voltage divider. The degree of hysteresis will depend directly on the voltage divider; if a 100-to-1 voltage divider is used, the applied input voltage will have to overcome a hysteresis of 1% of the output swing.

With a 100K resistor going from the output (pin 6) to the non-inverting input (pin 3), and a 1K resistor going from pin 3 to ground, only 1% of the voltage at the output is fed back to the non-inverting input. Because the 741 has a gain of over 100,000, it is easy to see why the output becomes saturated in either direction.

Anytime a sufficient voltage--of the same polarity as the output--is applied to the inverting input, the output will flip to the opposite state or polarity. In order for triggering to occur, this applied voltage must be higher than that found on the non-inverting input. For example, if the output voltage is plus 10 volts, the amount of voltage fed back to the non-inverting input will be about 0.1 volts. If just barely more than plus 0.1 volts is applied to the inverting input, the output will flip to its opposite state.

To make a circuit that will operate like a 555 configured as a VCO, a means of charging and discharging a timing capacitor at a rate that is proportional to the input voltage has to be devised. The timing capacitor goes from the inverting input to ground. The output goes through a diode to two resistors: One is a discharge resistor whose other end goes to the inverting input and to the capacitor. The other is the input resistor whose free end goes to the driving voltage. The orientation (direction) of the diode is what determines which polarity of input will charge the capacitor and cause the circuit to oscillate.

Circuit Using a Single Op-amp

Both plus and minus power supplies are required; their common connection is at circuit ground.

If a single op-amp is used (such as a 741), pins 1, 5 and 8 are not used. Pin 4 goes to the negative supply; pin 7 goes to the positive supply.

The inverting input (pin 2) goes through a timing capacitor to ground. (This cap can be of almost any value; I used 0.01uF for my test circuit.) Pin 2 also goes to a discharge resistor (R2). I used 100K here. The other end of (R2) goes through a 1N914 diode to the op-amp's output (pin 6 of the 741). The junction of the diode and R2 also goes to another 100K resistor (R1); the free end of R1 is the input of the VCO.

To create hysteresis: The non-inverting input (pin 3) goes through (R4) to ground (power supply common). Pin 3 also goes through R3 to the output (pin 6). I chose 1K for R4 and 100K for R3.

If you put the anode of the 1N914 onto the output and the cathode at the junction of R1 and R2, it will take a positive voltage at the open end of R1 to make the VCO work. With the cathode on the output and the anode at the junction of R1 and R2, it will take a negative voltage at the open end of R1 to make the VCO work.

If you change the value of either R3 or R4, you will change the voltage-to-frequency ratio of the circuit. Keeping the resistance ratio of R3 to R4 high is usually a good idea, since making R4 comparatively large will increase the threshold of hysteresis that the input voltage must overcome. (Of course, you can change the VCO sensitivity by choosing different resistors for R1 and R2, and by changing the value of the timing capacitor.)

Another thing you can do to this circuit is to apply an offset so as to put its "threshold of oscillation" precisely where you want it. The simplest way to do this is to take the bottom end of R4 (the end shown here as being grounded) to the wiper of a pot instead. The ends of this pot can go to plus and minus voltages as desired. Remember, though, that the resistance of this pot--and the relative position of its wiper--will figure into the hysteresis, so the pot will have to be of fairly low resistance value. (Bypassing the wiper won't really help, since while the oscillator is dormant, this bypass cap will charge in such a direction as to increase the threshold.) Use of another op-amp to buffer the voltage from this offset pot would solve that problem.

BULLETIN BOARD

Here's a man of action. He sent me this letter on an IBM diskette. I'll follow it with comments.

* * *

"One thing I would like our SK editor to do is to include print diagrams on all SK projects for those who so desire, even if this means paying extra to get them. Though I cannot see well enough to read drawings, they sure are a help when working with sighted friends and co-workers. Perhaps we could also get him to tell us about some of the latest research and developments for the blind going on in the SK labs.

"In addition, we need better ways of finding out who makes new, interesting and useful products for us.

"I am working on a cross-linked directory for all of the past SK Technical Files for those who want it. It will have such features as an alphabetical listing of all the files, the issues in which they are found, what they are about, a listing of subjects--what they are called and where they are found, etc. Other information that I may include would be: pin outs and technical data on commonly used IC's and transistors, and phone numbers and addresses of suppliers.

"I would also like to compile a list of other SKTF readers who could be contacted for help. (Maybe we won't have to pester Bill and Tom with our silly phone calls so much if we can help each other more.) I hope to be done with this challenge some time this year; if anyone else is working on such a project, please notify me.

"I will update this information once a year. I intend to turn it over to Bill and Tom for printing and distribution.

"I would like to help more by being a resource center for the blind electronics person. Outside of SKTF, the things I could help with are: being an information center for where to get the best prices and selections on parts and equipment, technical information and so on. If anyone else would like to help with these ideas (especially if you are an IBM computer user), please phone or write to me (no Braille please), or send a 1.2meg or 360K 5-1/4-inch floppy disk to me: Bernie Vinther, 915 West Grande Ronde, Kennewick, WA 99336; phone: (509) 586 8060."

* * *

These are laudable goals, and I marvel at your ambition. Even with computers, I notice that those jobs take longer than they seem. However, we wish you the best of luck. Some afternoon, maybe I can help by fishing through the Braille library and reading the abstracts. (I volunteered to help Lloyd Rasmussin at the National Library Service to do that kind of index for the "Braille Technical Press"; I haven't so much as gotten off the ground. Good luck to you.

As for surveying new products and writing them up, I just don't want to take up the pages--although it might put the magazine back on time, wouldn't it now). There are "resource guides" and places to go for much of that information. For example, "Playback Magazine" (on tape) has recordings of people doing off-the-cuff reviews of products they have found, and its editor, Ed Potter, reads card-files of merchants' phone numbers. I bought the most recent local Braille resource guide, which will be listed at the end of this discussion.

One thing I don't want to do is publish information for comparative shopping. I don't want to get too close to something that looks like advertising, since I have a non-profit funding institution and free mailing privileges to protect. (When I stuff fliers in the mag for you, I mean no endorsement of gadgets or products.) I have reviewed things now and then, but I can call that activity "research evaluation." A guy like Ed Potter at "Playback" doesn't use free mailing or non-profit status, so he can schedule some pretty lively contests.

What I don't mind doing is being "the list of lists." There are computer magazines and catalogs which I should list, and I'll try to attend to them.

The good news is that you can get yourself on the mailing list of the Smith-Kettlewell Rehabilitation Engineering Center's "Annual Report of Progress," where all of our research activities are delineated in excruciating detail. The bad news is that this is only in print. It has nifty things like reports of progress on Polaroid refraction for detection of potentially blinding impairments in infants.

As old-timers among you remember, we compiled a catalog of our projects for two years running (back in 1982-1983). For shame, we have not done that since; this has been strictly for lack of time and help. Any item of general interest, I make public herein; however, through the catalog you could have found out about the brake-drum micrometer we designed for the guy who could never use it, since California law said that anyone rebuilding auto brakes had to test-drive the car. Other neat things appeared in there, though; a notable example was the prototype "Auditory Arcade" which was a workbench-like toy for kids. We know we need a new catalog, so I have hopes for its renewed existence.

Where you really are in luck is on the subject of print schematic diagrams. Anything we develop has to evidence itself as fruits of research; in most cases, this means that a project must exist on paper so that it is reproducible. What's more, the diagrams are free, whether or not you subscribe to this fine magazoone. The catch is that production of the diagrams bears little relation to the schedule of SKTF--in fact, they are usually far behind. (Creation of print drawings coincides with the issuance of our annual reports.) If you write for a print circuit, we will put you on a list, and when it gets formally drawn, we'll send it to you.

What we haven't found time to do is draw diagrams of non-SK projects from descriptions in these pages. Thus, if you want a print schematic diagram of Bernie's "Little Blue Box," we can only send you to him with good wishes.

Quite a few circuits have gotten into these pages through blind inventors who haven't got a draftsperson at hand; it is a shame that those projects are not readily available to sighted people. (Nifty instruments like Mr. Britz's Multi-Microfarad Meter, and Albert Yeo's transistor tester remain out of sight, unfortunately.)

Yet there are plenty of technical people who could be asked for help. The retirees from the various telephone companies have an organization called "The Telephone Pioneers"; for years they serviced Talking Book machines. All major cities and many towns have ham radio clubs. These are largely service organizations looking for community projects. They could not only draw the diagrams, but they could--in learning how verbal diagrams work--serve blind folks as readers, and could build odd projects whose market is too small to interest businesses.

Results of the SKTF Survey done back in 1982 showed that of the 112 devices built out of these pages, 20.5% were built by people whose help was gotten on the outside. Go get 'em, boys--they're ripe for the picking (I think ... I hope).

* * *

A good resource guide for computer enthusiasts is "Resource Guide to Computer Access for Visually Impaired People," costing $15 and compiled by TRI Visual Services Corp., P.O. Box 8126, Sacramento, CA 95818; telephone (916) 428-8602.

EDITOR'S CORNER

Seasons Greetings! I never know quite what season it is, but I greet you all the same. (Actually, this lab is very dark all year round, and our main source of photons is a bank of soldering irons.)

Please excuse the delay. It is the result of a bad editorial decision--perhaps my first one? Our speech board, which you will see in Winter 1989, grew and grew. Modifications of it were still being made in April, and that could hardly be appropriate for this fallen Fall issue. I had every intention of including it this time, but Bernie Vinther bailed me out with his VCO.

The blood pressure gauges have been evaluated since the paper was written, and they seem to work respectably, according to medical personnel. While we were at it with the speech board, we used the National voltmeter chip to create a Spanish-language blood pressure meter for a conference in March (a March of a year following this issue). We will record it in French to be displayed in New Orleans.

* * *

Now that this issue terminates a year, please consider subscribing for 1989. I'd like to thank those who love this magazine so much as to subscribe already. Those of you who have paid for 1989 will notice that this shipment contains two issues.

Please note that 1989 flags a price increase. As of now, the Talking Book edition is $14, the diskette version is $16, and Braille stays the same at $18.

As I sit here trying to think of a clincher, the clock is ticking and I am making the magazine only later. Merry Christmas, and be grateful that I am not in charge of the International Dateline.