measuring stress, strain and vibrations

LAST month, prior to examininghumidity sensors, we discussedop.amps in relation to their offsetvoltages and bias currents and how thesecan affect the accuracy of sensor outputmeasurements.

This month we examine strain sensors,which typically produce extremely smallchanges in output signal. Such small sig-nals can easily be distorted by other exter-nal conditions. So it is appropriate to firstdiscuss various ways in which signal errorsand imperfections can be minimised. Wethen examine strain sensors themselves andalso look at piezoelectric vibration sensors.

Then, in Lab Work, we describe how tobuild a simple rain intensity meter, and anovel knocker circuit.

Sensors produce signals, that is varying

voltages or currents, whose variation car-ries information about whatever we aresensing or measuring. These signals, andthe signals within the circuits connected tothe sensors, can take a variety of forms andare subject to various types of error andimperfection.

Last month we looked at the problem ofoffsets, which are basically d.c. or very lowfrequency errors. At frequencies higherthan the slowly changing offsets, unwantedsignals are usually referred to as noise orinterference (see Figs. 4.1 to 4.4 for exam-ples of various forms of noise or error).This may comprise random variations inthe signal voltage or may have a very spe-cific frequency, such as 50Hz/60Hz mainshum, for example (Fig.4.4). We will look atrandom noise in more depth in a later part.

The signals we have just discussed are

called single-ended because a single wire(other than ground) is used to carry the sig-nal. A possible problem with this approachoccurs when the wire carrying the signalmay pick up noise, acting as an aerial andpicking up, say, radio frequency interfer-ence or mains hum.

To overcome this we can use a differen-tial signal, which is carried on two wires(i.e. two voltages V1 and V2) other than

128 Everyday Practical Electronics, February 2002

EPE Tutorial Series

TEACH-IN 2002Part Four Good Vibrations MeasuringStress, Strain and Vibration

Making Sense of the Real World: Electronics to Measure the Environment

IAN BELL AND DAVE CHESMORE

Fig.4.1. A signal varying around 0V.The peak value is 2V and the peak-to-peak value is 4V. The frequency is2.5kHz. There is no noise or error pre-sent, although the digitised samplingsteps are apparent.

Fig.4.2. The signal from Fig.4.1 with a1V offset. If you consider the signalrather than the offset the true peakvalue is still 2V, even though the signalpeaks at +1V and 3V.

Fig.4.3. The signal in Fig.4.1 withsuperimposed noise of a higher fre-quency than the signal.

Fig.4.4.The signal in Fig.4.1 with superimposed noise of a lower frequency than thesignal. This is about 0.2V peak of 50Hz mains hum. Note that the waveform is dis-played over a longer period than Fig.4.1 so that the noise is more obvious to see.

Fig.4.5. Differential signal. The signal isthe difference between V1 and V2 andtherefore has a peak value of 1V and apeak-to-peak value of 2V.

Fig.4.6. Differential signal with a com-mon mode noise. The differencebetween these waveforms is the sameas the signal in Fig.4.5.

Fig.4.7. Zooming in on the first part ofFig.4.6 to see more clearly that thenoise is common mode i.e. the errorgoes in the same direction on bothwaveforms.

Voltage/V

ground. Fig.4.5 shows a differential signalwith a peak voltage of 2V and a peak-to-peak voltage of 4V. Note that this is thedifference between V1 and V2.

With a differential signal, if the signal volt-age on one wire increases then the signal volt-age on the other wire decreases by exactly thesame amount. The actual signal is equal to thedifference in the voltages on the two wiresmeasured with respect to ground. So if thetwo voltages on the two wires are V1 and V2the signal Vs is (V1V2).

If the two wires run closely parallel, thenthe same error (e.g. mains hum, interfer-ence, etc.) will occur on each wire. If thiserror is (delta) then the voltage on wire 1will become V1 + and the voltage on wire2 will become V2 + . The signal is the dif-ference between the two wires, that is:

((V1 + ) (V2 + )) = (V1V2)which is the same as without the error! Thisis illustrated in Fig.4.6 and Fig.4.7.

The error voltage is common to bothhalves of the differential signal. It is there-fore called a common mode voltage andnoise of this form is called common modenoise. If the voltages on the two wires areV1 and V2 the common mode signal Vcm is(V1 + V2) / 2 (i.e. the average of the voltageon the two wires).

Differential signals quite often have d.c.common mode voltages, for example a 2Vpeak-to-peak sine wave differential voltagewith a 15V common mode d.c. signal isshown in Fig.4.8.

Differential signals are used quite com-monly in sensor systems. They have appli-cations such as temperature compensationand reducing the effects of interference ifsignals have to travel over relatively longwires. Closely spaced long wires carryinga differential signal will pick up interfer-ence, but this will influence both wiresequally and hence appear as a commonmode signal.

For temperature compensation wearrange two sensors so that only one is sub-ject to the condition we are measuring, butboth are affected in the same way by tem-perature changes. Alternatively, we can usetwo sensors, configured so that the outputof one will increase as other decreasesunder the influence of the measured quan-tity, but with both experiencing the sametemperature.

In both cases the unwanted changes insensor output due to temperature changesappear as a common mode signal, whereasthe measured quantity is differential.

A common approach to generating a dif-

ferential signal from a sensor, or sensorcombination, is to use a bridge circuit. Inits most basic form, the bridge circuit con-sists of two resistive potential dividers asshown in Fig.4.9.

The four sections of the bridge (shownhere as four resistors, R1 to R4) are gener-ally referred to as the arms of the bridge.The sensor or sensors may be in any one ormore of the arms. Bridges do not have to besimply resistive, we can make capacitiveand inductive bridges too.

A voltage, Vs, is applied across thebridge, in many cases this is simply thepower supply of the circuit, although it maybe some other voltage and may even be ana.c. signal. The output voltage is the differ-ence between the two potential dividervoltages.

In the simplest bridge circuit only resis-tor R1 is a sensor, the other resistors arefixed and typically all have the same value,equal to the nominal value of the sensor.

For temperature compensation, R1 is theactive sensor and R2 is another sensor iso-lated from the measurement quantity, but atthe same temperature as R1.

For some sensor systems, such as strainmeasurement, it is possible to arrange twosensors that have equal and oppositeresponses to the quantity being measured(i.e. one increases in resistance and theother decreases as the quantity changes).

In such cases we can build push-pullbridges in which either R1 and R2 areopposite sensors and R3 and R4 are fixed(single push-pull), or in which R1 and R4are equal (say) negative-response sensorsand R2 and R3 are equal (in this case) pos-itive-response sensors (double push-pull).Single push-pull provides twice the outputsignal of a simple bridge and double push-pull four times as much.

When connectingsensors or bridge cir-cuits to amplifiers (orother circuits) we needto be aware of the pos-sible problem of load-ing the sensor orbridge with the ampli-fier input. If a sensoroutputs a voltage wecan view it as an idealvoltage source (Vs) inseries with a resis-tance (Rs), called theinternal or sourceresistance. This is

connected to the amplifier input that has acertain input resistance (RI). This is illus-trated in Fig.4.10.

Observant readers will notice that RS andRI form a potential divider. Thus the volt-age at the amplifier input, taking loadinginto consideration, is given by:

V1 =RI VS (Equation 1)(RS + RI)

We get this equation by using OhmsLaw to get the current (Is) through the tworesistances (VS divided by the total resis-tance (Rs + RI) and applying Ohms Lawagain to get the voltage drop across RI (bymultiplying RI by this current).

From the equation we see that if we wantVI to be as large as possible then RI must bemuch larger than RS (we are assuming RS isfixed for a given sensor, but RI is influ-enced by our choice of amplifier). If RI isvery much larger than RS then the loadvoltage is effectively equal to the sourcevoltage.

It may seem that Fig.4.10 implies thatthe sensor generates the voltage signal.This is not necessarily the case, as we canrepresent a wide variety of circuit configu-rations in the form shown in this figure.The representation of a circuit as a voltagesource and a resistance is known as theThevenin equivalent circuit.

Lets look at finding an equivalent cir-cuit. Consider a sensor whose resistancechanges from 100k to 200k as thequantity being sensed varies (e.g. it couldbe a thermistor or a light dependent resis-tor, such as those used in Lab Works 1 and2). We could wire this sensor into a poten-tial divider connected to an amplifier asshown in Fig.4.11.

In an ideal circuit the potential dividervoltage might vary from 480V to 686V asthe sensor resistance varies. We willassume that the amplifier level shifts andamplifies this 206V variation to give an

Everyday Practical Electronics, February 2002 129

Fig.4.8. 2V peak-to-peak differential sig-nal with 1.5V common mode voltage.

I

Fig.4.9. Basic Bridge Circuit Fig.4.10. Thevenin equivalent circuit forsensor and amplifier configuration.

I

S

Fig.4.11. Sensor and amplifier.

output using the full 0V to 12V range of thesupply. The details of its implementationare not important, except to note that weknow the input resistance of the amplifieris RI.

The values in the equivalent circuitshown in Fig.4.12 vary depending on theresistance value, so we will just look atwhat happens when the sensor has a valueof 100k.

The value of Vs is simply the open circuitvoltage i.e. the output voltage with noload, which we have already mentioned.This is easy to calculate in this case, as it issimply the potential divider voltage. Whenthe sensor has a value of 100k this is480V as previously stated.

The value of RS is a little more difficult.It is calculated by taking the short circuitoutput current from the original circuit andfinding the resistance that would give thesame short circuit current with Vs.

The short circuit output current for ourpotential divider occurs when we connectthe potential divider point to ground. Thecurrent is 12V/150k = 80mA. To get80mA with 48V we need 48V/80mA =60k. So Rs = 60k when the sensor hasresistance 100k. Note that in this case thevalue of Rs is equal to the parallel combi-nation of RI and the sensor.

We can now draw the equivalent circuitas shown in Fig.4.12. From this, andusing the loading equation (Equation 1)from earlier, we can see effect of RI. If RIis, say, 5k we get 48 5/(60 + 5) =037V input to the amplifier rather thanthe 48V we would hope for. However, ifwe use an amplifier with an RI of 50Mwe get 479V at the amplifier input, pret-ty close to what we want. For other sensorresistance values we could perform simi-lar calculations.

In this example we could use the equiv-alent circuit with the voltage source eventhough the sensor itself does not generatea voltage. Electronics designers often useequivalent circuits. Using sets of rulesand often some approximations, theytransform a real circuit into a simplerequivalent that behaves in the same way(at least with respect to something theyare interested in).

Equivalent circuits contain fewer com-ponents (they are abstract rather than realcomponents) which makes subsequent cal-culations a lot easier. Furthermore, compar-isons between different circuitstransformed into the same equivalent cir-cuit are easy to make. The Thevenin equiv-alent circuit is good example of thisapproach as all sources are represented ina similar way.

Having highlighted how small signals

can be kept relatively free from errors andimperfections, we can now examine howsmall levels of stress and strain in materialscan be measured.

Measuring force effects on materials isone of the most common applications forsensors. For example, all weighingmachines rely on the weight of the object tobe determined to produce a force that canbe measured.

Pressure sensors operate in a similar man-ner, whereby the pressure difference betweenone side of a diaphragm and the other causesthe diaphragm to move and experience aforce. Accelerometers use changes in forcecaused by changes in speed.

In most cases, these forces causechanges in an elastic material, which inturn change its resistance or other proper-ties, the change can then be measured toobtain the force.

Lets look first at the effects of applying

a force to a solid object such as a bar ofmetal. Such a force is called stress but it isnot the same stress we get when our com-puter stops working yet again, although wemay very well want to apply considerableforce to the computer!

If we look at the constitution of a metalat the atomic level, it is a lattice of atomsheld together in equilibrium and the spac-ing between atoms determines the physicalsize of the object. When we apply a force,the atoms will re-arrange themselves inorder to keep equilibrium. The atomicspacing will change and so will the physi-cal dimensions, i.e. the object will deform.This resulting deformation is called astrain. Panel 4.1 and its Fig.4.13 give adetailed description of stress and strain.

So how do we create a sensor that can mea-

sure strain? As you may know, a length ofwire of any metal has a resistance (R0) whichdepends on its length (L0), cross-sectionalarea (A0) and resistivity ( in s) as follows:

R0 = L0 (Equation 2)A0

When the wire is stressed, its length willincrease by L and its area decreases becausethe overall volume of the wire remains con-stant. Without going through the mathematics,the approximate change in resistance due tothe length and area changes is:

R 2R0L (Equation 3)L0

This equation shows that the strain(L/L0) is converted directly into a change inresistance. A sensor based on this principle iscalled a strain gauge and is a length of metalwire or foil glued to the object whose strainis to be measured. When the object isdeformed so is the strain gauge.

A typical strain gauge is illustrated inFig.4.14 and consists of a wire or foilarranged in a specific pattern. The idea is to


Fig.4.12. Equivalent circuit for Fig.4.11when the sensor resistance is 100k.

PANEL 4.1. Stress and StrainIf we consider a metal rod of length L

metres (m) and cross-sectional area Asquare metres (m2) as shown in Fig.4.13and we apply a force F Newtons (N) insuch a way as to pull the rod apart, wecan define the tensile stress as

Tensile Stress = FAthe units are in Newtons per square metreNm2

The strain is defined as the change inlength due to the force divided by thelength, i.e. the fractional change inlength. The change in length is usuallywritten as L where means change. Sothe tensile strain is defined as:

Tensile Strain = LLthis has no units

Application of a force in the oppositedirection so as to push the rod togethergives compressional stress andcompressional strain. A third form,shear stress/strain is obtained when wetry to push the material sideways. Inthis case, the shear strain is defined as:

Shear Strain = XL

where X is the change in width and L isthe width of the object. If we plot on agraph the change in strain for variousvalues of stress we get a straight lineuntil the point where necking occurs, i.e.the material suddenly becomes softer andthinner. The material finally breaks ifstress is further increased.

The slope of the line is called the mod-ulus of elasticity or Youngs Modulus (E)and is equal to stress divided by strain.Different materials have different valuesof E. For example, aluminium has E =689 1010, copper 1173 1010 andpolyethylene (a form of plastic) 345 108.

Fig.4.13. Forces on a rod of metal.

Fig.4.14. Layout of a typical straingauge

cause the wire to change length by as muchas possible when a force is applied in onedirection only, i.e. it is unidirectional. Thepattern is designed so that the length of thestrain gauge is as long as possible.

Before we start using strain gauges incircuits there is one more thing to consid-er the gauge factor (GF). The GF of astrain gauge determines its operatingcharacteristics and is an accurate measureof the strain-resistance relationship wehave just described (Equation 3).Impurities in the metal, and indeed thetype of metal, lead to small deviationsfrom the ideal and the GF takes this intoaccount. GF is defined as:

GF = R/R (Equation 4)L/L

Gauge factor is always close to a value oftwo for metal strain gauges but can be ashigh as 10 for special alloys or gaugesmade of carbon. The larger the value of GFthe better, since we get a larger change inresistance for a given value of strain.Typical resistance values for commercialstrain gauges are 60, 120, 240, 500and 1000.

How much does the resistance of a straingauge change by? Strain values are usuallyvery small as it requires a lot of force tostretch a wire. A typical value might be0001, corresponding to a change of 1mm

in a length of one metre, and would lead toa change in resistance of only 024 for a120 strain gauge. Such small resistancechanges are not easy to measure and canlead to many difficulties.

As we have seen, the change in resis-

tance is small and if we pass a currentthrough the strain gauge, the resulting volt-age change will also be small. We need ahigh gain amplifier to increase the voltagelevel to a usable value.

Strain gauges are also temperature sensi-tive and require some form of temperaturecompensation. However, we can use adummy gauge placed close to the activegauge in such a way as to be insensitive tothe forces (see Fig.4.15).

The two gauges are placed in a bridgecircuit as shown in Fig.4.16. Any changesin temperature will be cancelled out. Wecan also increase the sensitivity in someapplications.

In next months Lab Work we showhow to build a crude weighing machinebased on the bending of a metal barwhere a gauge placed on the top willexperience tension and a second gaugeplaced underneath will experiencecompression. Using these in a bridgewill effectively give twice the outputvoltage, whilst also providing tempera-ture compensation.

Piezoelectric material produces a voltageif it experiences a force and also bends if avoltage is applied. This is the basis for allcrystals (as used in oscillators) and also forpiezoelectric sounders.

A crystal and a piezo-sounder have sim-ilar construction a thin slice of piezoelec-tric material placed between twoconducting plates forming a capacitor. Ifwe apply a force to one plate, we get a tran-sient voltage produced across the plates,and conversely if we apply a voltage acrossthe plates, the plates will bend.

A sounder operates by applying a rapid-ly alternating voltage. This causes theplates to bend backwards and forwards,moving the air adjacent to the plates andproducing a sound. Each sounder will havea resonant frequency at which the outputsound level is much stronger.

This resonance is used in crystal circuitsto produce highly stable and accurate fre-quencies. The magnitude of the piezoelec-tric effect varies depending on the materialused. One of the best is lead zirconiumtitanate (PZT).

Another piezoelectric device is known asa bimorph which consists of two layers ofpiezoelectric material separated by a thinlayer and with electrodes on the outsidelayers. If an electric field is applied across


Example of a dual strain gauge module.

Fig.4.15. Using a dummy strain gaugefor temperature compensation.

PANEL 4.2. Strain Gauges in UseStrain gauges are delicate components

that need to be handled very carefully.They usually have two very fine wiresand are supplied with miniature self-adhesive printed circuit boards carryingtwo copper tabs. Once a strain gauge hasbeen affixed to the item to be measured,the p.c.b. should be attached close to thegauge and the fine wires soldered to oneend of the tabs. Stronger wires can thenbe soldered to the other pads on thep.c.b., which can then be connected to themain measurement circuit.

Strain gauges must also be matchedto the metal, i.e. if we are using aluminiumthen we must use strain gauges designedto be matched to aluminium. This isbecause the strain gauge will have thesame temperature coefficient and willexpand at the same rate as the metal if thetemperature changes. The most commongauges are matched to steel or aluminium.

A strain gauge is attached to a metalsubstrate simply by gluing it. The adhesive

chosen should ideally be non-elastic (e.g.an epoxy adhesive) otherwise any strainexperienced by the metal will be absorbedby the adhesive and the strain gauge willnot register any changes.

The small terminal carrier p.c.b. mayalready be coated with an adhesive back-ing, so it can be applied nearby, alwaysensuring that the surface is completelyfree of grease and dirt. Should the p.c.b.fall away in use, this could eventuallycause damage to the strain gauge.

We have already mentioned that thechange in resistance, and therefore volt-age, is very small and must be amplified1,000 times or greater. This means thatthe circuit is susceptible to interferenceand the wires connecting the gauge to themeasurement circuit should be shielded,i.e. coaxial cable should be used. If this isnot available then the wires should betwisted together to help cancel any noise.We will offer a practical demonstration ofstrain gauges in next months Lab Work.

Fig.4.16. Strain gauge circuit.

Example of a simple piezo discsounder.

the electrodes, the bimorph will bend in asimilar way to a bimetallic strip in a ther-mostat. The bimorph will bend in the oppo-site way if the field is reversed.

Similarly, it will generate a voltage if it isbent and can be used as a vibration sensor.Inexpensive bimorph elements that arereadily available are 15mm long, 2mmwide and less than 1mm thick. They makevery good vibration sensors but are fragileand can break easily.

We have all heard of the Hubble SpaceTelescope which is in orbit around the earthand seen the spectacular images it pro-duces. The reason for it being above theatmosphere is that it is immune fromatmospheric effects such as diffraction andturbulence.

Recently, large telescopes have beenbuilt that use a new technique known asadaptive optics which attempts to removethese effects by continuously deformingthe mirror by small amounts. This could bedone by solenoids but their reaction time islong they and cannot operate at high

speeds. Bimorph elements can and havebeen used successfully in this application.

There are some other very interestingapplications of bimorphs which includeminiature actuators for tiny robots (walkinglegs) and developing cochlear implants fordeaf people where the bimorph convertselectrical energy from a microphone intomovement within the ear. Bimorphs arealso used when precise very small(micrometres) movements are needed forpositioning.

One of the most unusual applications ofpiezoelectric material is to sense gaseouschemicals such as carbon dioxide and sul-phur dioxide. These sensors rely on the factthat thin slices of piezoelectric materialwill have a resonant frequency dependingon the area of the slice and its thickness,which is the principle behind crystals asused for timing purposes in oscillators.

If a crystals enclosure is cut open, a circleof material with electrodes deposited oneither side will be seen. This construction isturned into a chemical sensor by placing a

thin layer of a chemical on one side of thecrystal that will react with, or absorb thechemical we are looking for. If the measur-and (substance to be measured) is absorbedonto the crystal, the crystals mass willincrease and its resonant frequency will drop.

Similarly, if the chemicals react, again themass and frequency will change. Such sen-sors can be very sensitive and are capable ofmeasuring in parts per billion (1,000 mil-lion) (micrograms) but suffer from onemajor drawback the reaction may not bereversible, i.e. the sensor can only be usedonce, or until all the chemicals have reacted.

There are many other sensors capable ofmeasuring forces (such as force sensingresistors) but in this part of Teach-In we areonly looking at piezoelectric sensors andstrain gauges. In Lab Work we show howwe can use commonly available piezoelec-tric sounders to measure forces such asvibration, by building a novel rain detectorand a doorbell that is activated by knockingon the door three times!

We will also be looking in future instal-ments at other sensors that use straingauges and piezoelectric materials such aspressure sensors.


TEACH-IN 2002 Lab Work 4ALAN WINSTANLEY

Detecting Vibrations

LOUDSPEAKERS rely for their principleof operation on the fact that applying asignal voltage across them produces amovement in the speaker cone ordiaphragm. This causes an air displacementthat, depending on its frequency, isdetectable as a sound wave by the ear.Different types of loudspeaker are designedto cope with high frequencies (tweeters), orthe long throw needed for bass notes(woofers) or frequencies somewhere in themiddle (mid-range loudspeakers).

Compare this with the principle of themicrophone: it is in effect a loudspeaker inreverse: sound waves impinge on themicrophone diaphragm or element, whichcould be a crystal, a static-charged electretmembrane or a dynamic moving coil.The movement of the microphone elementconsequently generates a voltage acrossthe microphones terminals that can thenbe amplified.

This months practical Labs use a formof loudspeaker (or sounder) as an effectiveelectronic method for detecting vibration.We provide a couple of application circuitsthat could readily be adapted to otherapplications including monitoring andsecurity applications.

A simple piezo disc element can beutilised as a form of microphone, which willgenerate a tiny voltage when it detects animpact, vibration or other pressure wave.

Two types of piezo sounder are readilyavailable. A simple piezoelectric disc is

nothing more than the sound element itself.To utilise one as a sounder or alarm tonegenerator, a separate oscillator driver cir-cuit is necessary. It should not to be con-fused with a self-contained piezo sounderwhich already has the necessary tone gen-erator incorporated into a plastic enclosureand, therefore, only a supply voltage isrequired.

The efficiency of modern piezo alarms isextremely high, characterised by highoutput levels for a relatively low powerconsumption; even some small units cancreate ear-splitting sounds.

For the following Labs, we will be usinga plain piezo disc which does not have atone generator.

Lab 4.1 Rain Intensity MeterIt is easy enough to detect the presence

of rain, and even the amount of rainfall, buthow about discriminating between a lightshower and a torrential downpour? A goodway of doing this is to sense the impact ofthe raindrops themselves.

Lab 4.1 will detect such vibration or aseries of impacts, and it produces a d.c.voltage that is proportional to the intensityof the raindrops. A heavy shower outputs ahigher voltage. It is extremely sensitiveand the circuit lends itself for use in otherapplications.

We rely on the impact of rain drops on apiezoelectric disc (which is actuallyaffixed underneath a plastic or metal plate)to produce a short voltage pulse. Fig.4.17shows the complete circuit, which consistsof two op.amps, the first of which, IC1a,acts as a non-inverting amplifier with again between 35 and 155. This can be setby VR1 to allow different sounders to beaccommodated.

The output of IC1a is connected to apeak detector circuit built around diodeD1, resistor R4, preset VR2 and capacitorC1. This is a basic form of sample and holdcircuit and its operation is straightforward.

Imagine a positive pulse being generatedby IC1a, therefore diode D1 will be for-ward biased and so capacitor C1 willcharge rapidly to the voltage of the pulse(minus 06V for the diode forward voltagedrop). After the amplifier pulse has

Examples of piezo sounder elements.The outer two need an external oscil-lating signal. The centre one is self-contained with its own oscillator circuit.


returned to zero, the voltage on the capaci-tor will slowly fall as it discharges to 0Vthrough R4 and VR2.

The time taken to fall will depend on thesetting of VR2 and can be varied fromapproximately 5 seconds to 15 seconds.The period can be made longer by increas-ing the value of C1, to suit other applica-tions. Multiple pulses caused by rainfallwill keep the capacitor charged and thevoltage will remain at a given level.

The second op.amp, IC1b, acts as aunity gain buffer to ensure the peak detec-tor circuit is not unduly loaded.

The output of the circuit is a d.c. voltagethat represents the intensity of the rain, butthe principle is quite crude since manylight drops of rain or a single heavy dropmay give the same reading. Also, since thepiezoelectric sounder is in effect a capaci-tor, it will take time to respond and at veryhigh rainfall rate the output of the sounderwill be continuous.

This circuit is therefore only a simpleindicator of rainfall rate, but could easilybe expanded to make a comprehensivedetector and logger, especially if the outputis connected to an input of a microcon-troller such as a PIC.

The circuit is straightforward to build ona solderless breadboard. It can be poweredusing the Teach-In 2000 power supply toprovide 5V d.c. For convenience, a TL082was utilised in our experiments, but youcould try any dual FET input type.

Different sounders have been tried andthe best seems to be a large area sounder,such as the KPS-100 piezoelectric sounder.Virtually any sounder will work however,but remember to use a naked piezoelec-tric disc and not a self-contained buzzer.

Use trimmer preset VR1 to adjust forsensitivity and VR2 to control the decay ofthe signal.

The Picoscope ADC-40 can be used to

measure the output directly at pin 7 ofIC1b. Select a relatively slow timebase,e.g. five seconds per division. You candemonstrate this by directly inserting theleadouts of the piezo disc into the bread-board and resting the disc on the tabletop.

By experimenting with the gain, wefound the circuit would easily register afinger tapping the tabletop or fingers drum-ming nearby. The screenshots (next page)show the Picoscope waveforms measured

during the experiment. Notice how thevoltage rose and decayed slowly as the rateof impacts changed.

You could try connecting the output tothe Picoscope data logger and log theintensity of rainfall over, say 24 hours.Only allow the sensor to be impacted byrain, everything else must be kept absolute-ly dry.

If you decide to build the circuit into aworking project, we suggest gluing thesounder to the back of a plastic or a metalbox to increase the surface area andincrease the chances of capturing rain-drops. The box should be placed at a slightangle to allow rain to flow off the surface.

It is also a neat idea to place the cir-cuit inside the box to form a completely

Fig.4.17. Complete circuit diagram for the piezo-disc rain sensor.

N.B. some componentsare repeated between LabWorks

Lab 4.1Resistors

R1 560kR2 39kR3 100kR4 470k

All 025W 5% carbon film.

PotentiometersVR1 470k sub-min preset,

horizVR2 1M sub-min preset, horiz

CapacitorC1 10 radial elect. 16V

SemiconductorsD1 1N4148 silicon diodeIC1 TL082 or similar dual

op.amp

MiscellaneousX1 KPS-100 50mm piezo-

electric speaker

Lab 4.2Resistors

R1 560kR2 39kR3 470kR4 39kR5 4M7R6 3M3R7 56k

CapacitorsC1 1n to 100n ceramic

(see text)C2, C3 1 tantalum or radial

elect. 16V (2 off)

SemiconductorsIC1 OP177, CA3140 or similar

FET input op.ampIC2 4093 quad Schmitt

trigger NAND gateIC3 4520 dual counter/dividerIC4 4098 dual monostable

MiscellaneousX1, X2 KPS-100 piezoelectric

speaker (2 off)

Approx. CostGuidance Only 1144

SeeSSHHOOPPTTAALLKKppaaggee

Breadboard assembly for the circuit in Fig.4.17. The sensor is not shown, but itsleads can be seen at the top left.

self-contained vibration or impact sen-sor, preferably using coaxial cable toconnect to other control circuitry asrequired. Use silicone sealant to seal anycable exits.

Can you think of any further applicationsfor a completely solid-state, self-containedimpact detector? How about a burglaralarm that detects an impact on glass, doorsor floors? By setting the sample and holdcomponent values accordingly, you couldcater for false alarms caused by minorimpacts. Perhaps it could be incorporatedinto vehicle alarms as well.

More experienced readers will be able toadd on a variety of simple circuits, perhapsbased on a Schmitt trigger or a thyristor,that could be triggered by the output volt-age of IC1b.

Lab Work 4.2 Door Knocker CircuitThe circuit for Lab 4.2 is a bit of fun but

could be incorporated into a novel doorbellwhich is activated by a number of knocks(say three) on the door. Fig.4.18 shows thecircuit diagram and the first thing to noticeis that the knock sensor based aroundop.amp IC1 is virtually the same as the sen-sor for rain intensity described in the previ-ous experiment.

The only difference is that the gain isfixed at about 13 to ensure we get a goodpulse out when someone knocks thesounder. There is plenty of scope for exper-imentation; we tested a variety of differentop.amps including the OP177 and theCA3140. The latter has MOSFET inputsand bipolar outputs. Be prepared to experi-ment with the amplifier gain values toobtain the best results in your own bread-board experiments.

The rest of the circuit may seem complexbut it is quite simple to analyse. Fig.4.19shows the timing diagram which shouldmake the circuits operation clearer. We startwith the output of the amplifier (point A),

which is a short pulsewhen the sounder isknocked. This pulse ismade longer (point B)

by the low pass filter formed around the RC(resistor-capacitor) network R4 and C1.

Gate IC2a is a Schmitt NAND gate con-nected as an inverter to form a clean pulse(point C). This pulse is input simultaneouslyinto counter IC3 and one half of a 4098 dualmonostable, IC4a. The latter is connected asa non-retriggerable monostable by connect-ing its Q1 output (pin 6) to its +TR2 input(pin 4). This stops any more pulses restartingthe timing pulse.

The Q1 output of monostable IC4a will behigh for two seconds when triggered. Theinverted output, Q1 (pin 7) is connected to thereset input (RST1) of counter IC3 to enable itfor a period of two seconds. The idea here isto count knocks for two seconds only andthere must be (say) three knocks or more forthe doorbell to be activated.

Since the counter is disabled for the first

knock, it will count to two for three knocks.Between them, gates IC2b and IC2c detectthis and the output of IC2c (point D) trig-gers a second monostable, IC4b, for aboutone second.

The output of IC4b (pin 10) could be usedto activate a relay to sound the doorbell. Here,though, we have illustrated the principle ofoperation by turning on a piezo sounder (X2)for about a second. NAND gate IC2d is con-nected as a modest audio oscillator which isenabled by the Q2 output of IC4b.

An important point to note is the fact thatthe second monostable will be triggeredonly if the count is three or four and onlywhen the two second interval has passed.You can, however, vary the values byaltering the monostable timing components(R6 and C3) to suit.


Picoscope displayof rainfall monitoredby the circuit inFig.4.17 at the out-put of IC1a.

Smoothed rainfall monitored at IC1b.

Fig.4.18. Complete circuit diagram for the knock three times sensor.

It is relatively straightforward to buildthis circuit on a long solderless breadboard,provided you work methodically throughthe circuit. Use fine long-nose pliers asnecessary to help insert wires.

A pair of KPS-100 piezoelectricsounders were used both for the knock

sensor and the sound element, with bothelements resting on the table alongsidethe breadboard. It was found that by giv-ing the table several sharp raps in succes-sion underneath the piezo discmicrophone, the sounder operated for ashort period.

Do, however, be prepared to experimentwith the amplifier, as this is the most criti-cal aspect in ensuring that the circuit func-tions successfully. If necessary, monitor thecircuit with a logic probe or your Picoscopeto check what happens when the piezo discX1 is subject to a tap nearby. (It is best notto knock the quite delicate piezo elementitself.)

The best approach to practical design isto put the outdoor knock sensor behind aplate which has the words on it Knockhere 3 times to ring the doorbell. We aresure there are many other applications ofthis simple circuit, such as detecting some-thing thrown at it you could make athrowing game similar to a coconut shy andring bells or turn on lights if a ball hits thecoconut.

We hope you enjoyed learning aboutsome more unusual uses for piezoelectricelements. With a more serious applicationin mind, the circuit values could readily beadapted to warn of a series of impacts orexcessive vibrations, e.g. to detect a break-in, as the need to have a rapid succession ofimpacts before the sounder is triggeredmay help prevent false alarms.

Next month we offer a sensitive strain

gauge circuit which uses some of the prin-ciples outlined in this tutorial section. Wewill also be moving on to describe instru-mentation amplifiers, and in forthcomingparts we will be investigating the principlesbehind the detection of acceleration andpressure.


Fig.4.19. Timing waveforms for the circuit in Fig.4.18.

Above: Breadboard layout for the circuit in Fig.4.18. Below: Sensors connected tothe above layout.

Single knock waveform at Fig.4.18point D.

Triple knock waveform at Fig.4.18point A.

Multiple knock waveform at Fig.4.18point C.

measuring stress, strain and vibrations

Documents