NEWLOW-NOISETRANSISTORCIRCUITFOR

ELECTROSTATICMICROPHONES

By P. J. BAXANDALL,B.Sc.(Eng.I'1

Amplitude-modulated R.F. Bridge Method with Many Advantages

THE conventional way of using a~ electrostatic (orcondenser) microphone is shown, in its simplestform, in Fig. 1. The resistance R is made so largethat, even at low audio frequencies, insufficient cur-rent can flow into or out of the microphone capacit-ance C, during one audio cycle, to cause a significantalteration in the stored charge Q. Since Q=CV, itfollows that if Q is k~pt constant, the voltage V acrossthe capacitance must vary when acoustic pressurecauses C to vary. With the values shown, theresponse will be 3 dB down at about 30 cis. Fromthe point of view of signal-to-noise ratio, however,it is advantageous to use an even higher value of

+

MICROPHONE

(SOp) f:.LkL~~

.,.

: IOOV..L.

Fig. I. Conventionalelectrostatic micro-phone circuit.

OUT

looMsR

E

by the required low-resistance than that dictatedfrequency response.

When, in 1957, the writer first considered the prob-lem of using an electrostatic microphone with purelytransistor circuitry, it was quite obvious that the im-pedances involved in a circuit of the Fig. 1 type werefar tOOhigh for it to be practicable simply to replacethe valv~ by a transistor. *

However, by operating the electrostatic micro-phone element in a radio-frequency circuit, so thatits capa-:itance variations are caused to modulate anrJ. carrier, the above-mentioned high impedances

--.---------

~ It i, intere5ting to reflect, however, that the notion of tran-5istorizing the Fig. 1 circuit now seems to be much more nearlya satisfactory practical proposition than it did in 1957. This isbecause some types of silicon planar transistor are now availablewhich will operate ;atisfactorily, in very high impedance circuits,at collector current, of a small fraction of a microamp.

Whilst the signal-to-noise ratio obtainable when using such atransistor in the Fig. 1 type of circuit would. probably be rather'.:>f.,rior to b~~atgiven by ~ valve, there are signs that other ampli-t;..".,,>-,,:devices may in due cours., become available which willl1''''.fI:OI1lC this limitation, One stich device is the insulated-gatenrid effect transistor (ref. 1) and another is the insulator valve(ref. 2).

538

are completely avoided and a very good performance ,;can then be obtained with semiconductor circuits. '

The general idea of using radio-frequency circuitsfor electrostatic microphones is, of course, quite old, ;

and both frequency modulation and amplitude modu- .lation have been employed.

F.m. systems have the disadvantage that random,noise f.m. on the oscillator output inevitably gives rise'to nois,:: at the audio output terminals. Since thewanted f.m. is usually of quite small deviation, thisnoise f.m. can prevent the overall noise performancefrom b~ing up to the highest professional standards.

In an a.m. syslem, however, by using a balancedbridge circuit, random noise modulation of the oscil-lator may be prevented from reaching the audio 'out-put terminals, and it was mainly for this reason thatthe author rejected f.m. systems right at the begin- '

ning and concentrated on a.m. bridge circuits-andif a bridge was to be used, then there was everythingto be said for employing the transformer ratio armprinciple first proposed by A. D. Blumlein.

R.F. Bridge Circuit

The broad outline of the system adopted is, then, :

to have a radio-frequency oscillator with a centre-tapped output winding, the microphone element anda capacitor of equal value being connected in series;across this winding, forming a bridge network. An.rJ. out-of-balance voltage is then obtained between'the junction of the capacitances and the winding'centre lap, of magnitude dependent on variations in:the microphone capacitance with acoustic pressure. .

This amplitude-modulated rJ. voltage is subse-.quently demodulated to recover the wanted audio.isignal. :

In the first experiments, the centre tap of the oscil-I~-'

lator winding was earthed and the bridge output wasituned to parallel resonance by an inductor to earth jfrom the junction of the capacitances. This output!was fed straight to a diode detector, the bridge being!set slightly out of balance to give some carrier output .and thus ensure linear demodulation. Quite encour-aging results were obtained, though it was found im-.

portant to select the right type of diode if excessive'detector-circuit noise was to be avoided. Ordinarypoint-contact diodes were hopelessly noisy, butG.E.C. EW78 silicon junction diodes (now obsolete)gave consistently good results (10 samples tried), the .

noise output then being only slightly above the ther-.mal noise level.

It was soon realized, however, that by employing ~

a proper phase-sensitive detector and operating the.~

WIRELESS WORLD, NOVEMBER 1963

Fig.2.system:-plete 'circuit. (diagraJprinciple.

bridge:term drand theforman<

It waof shunting tralphase-sibe madlthe neephone.ance wementsdisconti:

Fig. 2cui t finafully delof Britis

StartiJtransistcas beingabout 5using clbeen imbeen reqconsiderhelp wh.small mi

The cso as tohalves altwo end~two hahbridge, tmer forn

If thechange iJ

WIRELESS

JR

I '

B.Sc.{Eng.1

odperformanceor circuits.luency circuits.lrse, quite old,lplitUde modu-

e that random jitably gives rise:~ls. Since theideviation, this i

se performanceional standards.

Iing a balancedon of the oscil-]the audioout-jthis reason that.

t at the begin-1

e circuits-and'was everything irmer ratio arm '

nlein.

[opted is, then,with a centre-

.ne element andnected in seriese network. An.tained between.d the winding:on variations in I

oustic pressure. :

tage is subse-= wanted audio, ,

tap of the oscil-idge output was'lductor to earth:s. This outputthe bridge beingIe carrier output'. Quite encour-it was found im-..iode if excessive',ided. Ordinaryssly noisy, buts (now obsolete)mples tried), the:, above the ther-,

at by employingld opera ting the'

NOVEMBER 1963 '

IMc/s OSCILLATOR

+6V2sV r. m,s.

I

Fig. 2. The author'ssystem :-( a) Com-plete experimentalcircuit. (b) Simplifieddiagram to showprinciple.

(d)

(b)

PHASE-SENSITIVE RECTIFIER R.F.FILTER

REFERENCE VOLTAGE

ImH

Io= 600 fc =IOOkc/s

VI '-;T: I I rlosm",,,I "':::- V,~ 00i r I' J. 1

I ILLJL-- - - - - - - - - -- - - - - - - - - - - - _.JREFERENCE

bridge in a nominally balanced state, larger long-term drifts in the bridge balance could be toleratedand the possibility of degradation of the noise per-formance by oscillator noise would be reduced.

It was further realized that by using series insteadof shunt tUning of the bridge output, and by employ-ing transistors as low-impedance switches in thephase-sensitive rectifier, the output impedance couldbe made low (e.g., 600 ohms), and balanced, withoutthe need for an audio transformer in the micro-phone. Also it was t:xpected that the noise perform-ance would be excellent For these reasons experi-ments on circuits using diode detectors werediscontinued.

Fig. 2(a) shows the essential features of the cir-cuit finally adopted. This circuit was first success-fully demonstrated in July, 1959, and is the subjectof British Patem Application No. 6118/61.

Starting at the left-hand side, there is a single-transistor I-Mc/s oscillator. This circuit was chosenas being th~ simplest that would do the jub. It takesabout 5.mA at 6 volts, 'and operates in class B. Byusing class C operation, the efficiency could havebeen improved, but an extra capacitor would havebeen required in the emitter circuit-and one of theconsiderations is that every component saved is ahelp when it comes to building the circuit inside asmall microphone casing.

The output winding ,of the oscilhitor is bifilar,so as to obtain very tight coupling between the twohalves and thus to ensure that the voltages at thetwo ends will be very accurately in antiphase. Thetwo halves of this winding form two arms of abridge, the n:ticrophone and an air-dielectric trim-mer forming the other two arms.

If the bridge is slightly unbalanced, owing to achange in microphone capacitance, a small 1 Mc/s

WIRELESS WORLD, NOVEMBER 1963

11"2

-TT"

..

sine-wave voltage will appear at the junction of thecapacitances, and will have a magnitude propor-tional to the change in microphone capacitance.The phase of the voltage will change by 1800 asthe bridge swings through the balanced condition.Thus, assuming the bridge to be perfectly balancedinitially, the output waveform will be that of a sup-pressed carrier radio transmission when the micro-phone is acted upon by sound waves.

A very important point is that, looking back intothe bridge output, the above modulated r.f. wave-form comes from a source of quite low internalimpedance, i.e., the reactance of the two capaci-tances in parallel, which is about 1,500 ohms-verydifferent from the values of many megohms asso-ciated with conventional circuits.

Advantages of Tuning the Bridge Output:-Byseries tuning the bridge output by means of theinductor shown, the impedance seen looking intothe right-hand terminal of the inductor is madeeven lower-Q times lower, in fact-but the bridgeoutput e.m.f. is the same as before. Now, for agiven e.m.f., the lower the internal impedance ofthe source of the e.m.f., the greater is the availablepower. The fact that in this system the tunedbridge, regarded as a source of modulated d. outputsignal, has such a low internal impedance, is themain, reason for the excellent signal-to-noise ratioobtainable.

Of course, if there were no resistive losses, thatis if the Q were infinite, the internal impedance ofthe tuned bridge would become zero, and infinitesignal power would theoretically be available, atleast for very slow changes in microphonecapacitance.

In a practical microphone system the Q of the

539

series tuned circuit must not be made too high, other-wise the response of the system at high audio fre-quencies will be reduced, owing to sideband cutting,just as in a radio receiver. The resistor shown inseries with the tuning inductor limits the Q to anappropriate value, in the region of 15.*

The rest of the circuit is concerned with thedemodulating process, which is carried out by asimple phase-sensitive rectifier employing twojunction transistors.

These transistors are used simply as on-offswitches, which are operated by a reference voltagederived from the oscillator and fed in between theirbases and emitters through the transformer shown.\X'hen a transistor is driven" on" at its base, itbecomes capable of passing current in either direc-tion between emitter arid collector, or, in otherwords, it can function as a bidi1'ectiona[ switch.This is a great advantage possessed by transistors,as compared with valves.

Thus the two transistors, driven alternately intoconduction by the 1 II/leis reference voltage, per-

(d.)

~(b)

; Fig. 3. Phase-sensitive-rectifier emitter-voltage waveforms:-(a) 1.5 V d.c. applied to output terminals. (b) 20 kc/s sine-wave voltage applied to output terminals.

form the same function as the two-way switchshown in the simplified diagram of Fig. 2(b).

Consider one instant of time at which current isflowing from left to right in the inductor of Fig.2(b), the switch being supposed, at this instant, tobe in the position shown. Then, while this con-dition holds, the tendency will be for the top plateof the top reservoir capacitor to be charged posi-tively. During the next half cycle current will beflowing from right to left, but the switch will havechanged over to the lower contact, so that the ten-dency will now be to charge the lower plate of thelower reservoir capacitor negatively, and so on.

...---.-.--.-- .. __n__...._-------

* If the microphone amplifier input impedance is high com-pared with the output impedance of the microphone, then theresistor may be omitted without los~ of h.f. response and with animprovement in signal-to-noise ratio. The d.c. input resistance ofthe amplifier is likely to be almost zero, however, owing to theinput transformer, and if the damping resistor is omitted, a very~mall amount of unbalance of the bridge will gi:ve a large rectifiedcurrent. For this reason it is considered better to retain thedamping resistor even if an amplifier with a high a.c. inputimpedance is used.

540

Thus, all the time, the action of the circuit wiil beto tend to make the top output terminal positivewith respect to the bottom one. It is easy to seethat, if the bridge is' unbalanced in the oppositedirection, giving 1800 difference in the phasing ofthe inductor current with respect to the operationof the switch, then the opposite polarity of d.c.output is produced.

Some Practical Points:-During most of the experi-mental work the circuit was exactly as shown inFig. 2(a). No special arrangements were made foradjusting the phasing of the signal and referencein the phase-sensitive rectifier, though a slightphase adjustment is available by slightly detuningthl:' series tuned cir-:uit.

Later on, to improve 'the linearity of the demodu-lation process, the drive voltage to the base of eachof the switching transistors was increased by abouta factor of two, up to 3.5V r.m.s. This exceedsthe base-to-emitter voltage rating of the transistorsused, so two miniature point-contact diodes wereadded to prevent driving the bases too far positive.Small capacitors were shunted across the base

.resistors, now 4.7kn, to give a small reference-phasecorrection, thus allowing the series tuned circuit tobe set exactly at series resonance. These measuresimproved the linearity at the expense of a smallloss of signal-to-noise ratio. The measured resultsgiven later in this article were obtained with thesemodifications present, but the simpler arrangementis thought more appropriate for general use.

It may well be asked why the oscillator frequencywas made 1 Mc/s, and several considerations were,in fact, involved. The frequency must be highenough to give a good noise performance and a con-veniently low output impedance. A high frequencyalso makes rJ. filtering easier-the filter must havenegligible attenuation at the highest audio frequencyand 100 dB or so at the carrier frequency. On theother hand, the higher the carrier frequency themore difficult it becomes to get a really clean per-formance from the switching transistors. One hasa natural bias towards round numbers and 1 Mc/sseems about as good a choice as can be made.

The procedure adopted for setting the circuit upcorrectly is the foHowing. A 0-1 mA meter isconnected across toe phase-sensitive rectifier output,and the bridge is set slightly unbalanced to give asmall reading on this meter. The slug of the seriestuning inductor is then adjusted for a maximummilliammeter reading. Finally the bridge isbalanced for zero reading.

Sensitivity of Microphone Circuit:-With referenceto Fig. 2(b), the riO-load d. output voltage of thebridge is given by:-

.\

.\ VI SCV.=. ><--

- 2 C

where ~C is the amount by which the microphonecapacitance departs from its balanced value.

With no audio load on the final output terminals,no power can be supplied to the input of the phase~.sensitive rectifier, sir.ce there is nowhere for it togo. Consequently V" must be such that the peakvalue of the fundamental component of the square

.\

wave on the switch is equal to V2, thus giving zero

WIRELESS WORLD, NOVEMBER 1963

~ ,-.-

current

~_.... In thaudio 0capacit~

Audio (sider wilookingsensitivehow mlas a resoutPUt 1currentfrequen,

With2(b», thswitch VIVdr. <A

cui!, onl;wave WI

. flow in tfundametimes the

Thus'series tOJin the Sfsuppliedterminalspower su

'ing Vd.. Iresistancehence th(gives the

I

I

where [Zaudio fre

.; resistanceAt high

. plicated,separatedcircuit is

: tuned circby the ser

. total outpr.,. .t h'lr in ~

~f this in,audio freqtuned circ

(1)

current in Land 1'. This leads to the result:-

V117 SCV3=-4-X C-", ... ... (2)

In this equation V,. may be regarded as the peakaudio output e.mJ., hC being the peak value of thecapacitance variation.

Audio Output Impedance:-It is interesting to con-sider what will be [he audio output impedance seenlooking back into the output terminals of the phase-sensitive rectifier. All we need to do is to determinehow much direct current flows in the output leadsas a result of applying a direct voltage, Vd,., to theoutput terminals. The ratio of the voltage to thecurrent will be the output impedance, at low audiofrequencies at least.

With Vtlo betw~en the two switch contacts (Fig.2(b)), the waveform on the moving contact of theswitch will be a quare wave of peak-to-peak valueVtle. Owing to the selectivity of the d. tuned cir-cuit, only the fundamental component of this squarewave will be sigmficant in causing d. current to

. flow in the tuned circuit, and the peak value of thefundamental component of a square wave is 4/..times the peak value of the square wave itself.

Thus we can calculate the current flowing in theseries tuned circuit, and the power dissipated by itin the series loss n'slstance. This power must besupplied by the d.l:. source connected to the outputterminals, and :here is nowhere else where thepower supplied can be dissipated. Thus, by equat-

, ing Vtic.Lie to the power dissipated in the series loss. resistance of the tuned circuit, we may find Id,.and

hence the output impedance. Doing this in detailgives the result:-

I z""j 1.10'= ~\... .,. ... (3)

where [Z"ut]L1"is the output impedance at lowaudio frequencies and r is the total series lossresistance of the tuned circuit.

At higher audio frequencies things are more com-plicated, because the sidebands are then wellseparated from the frequency to which the tunedcircuit is tuned (1 Mc/s), and the current in thetUned circuit is ;.tffected by its reactance as well asby the series loss re5istance. Allowing for this, thetotal output imp~dance looks like a resistance of!. ..~r in series with' an inductance; the reactanceof this inductance is equal to the resistance at anaudio frequency .~qual to half the bandwidth of thetUned circuit. The inductive component is fairh'

negligible, even at 15 kc/s, in the design adopted,owing to the low Q of the tuned circuit.

Provided sufficiently fast transistors are used inthe phase-sensitive rectifier, the measured sensi-tivity and output impedance agree quite closely withthe calculated values. Semiconductors Ltd. surface-barrier transistOrs, type SB240, were chosen. OC44swere used in the earliest experiments, and whereasthese did produce re;-:~ults,the waveforms were farfrom the simple theoretical ones which would beproduced by an ideal switch, and the output im-pedance was considerably lower than the calculatedvalue.

The photographed waveforms shown in Fig. 3show that quite fast switching action occurs. Forthese waveforms ~he tuned circuit was disconnectedfrom the input terminal of the phase-sensitive recti-fier and a high-speed oscilloscope was connected rothis input point. The top waveform, a 1 Mc/ssquare wave, was obtained with a 1.5 V dry cellconnected to the output terminals of the phase-sensitive rectifier. For the lower waveform, the drycell was replaced by a 20kc/s sine wave from anoscillator.

Low-pass Filter:-Referring to Fig. 2(a) again, it willbe seen that a Jow-pass filtSr is included betweenthe phase-sensitive rectifier and the outgoingmicrophone line. This is to prevent 1'.f. currentsgetting out onto ~ile microphone cable, and to pre-vent d. signals from elsewhere, picked up by thecable for example, getting back into the microphonecircuits. This filter is very necessary, as otherwiseobjectionable heterodyne whistles could be genera-ted under some conditions. The design of the filteris, however, very uncritical-it must have little effecton the audio-frequency response, but must have avery large attenuation at 1 Mc/s and above. Thecut-off frequency ha~ been made 100 kc/s, and noclose-tolerance ,_omronents are required. Theattenuation at 1 Mc/s is about 100 dB, which iscomfortably sufficient. The inductOrs were woundon t in outside diameter ferrite toroids, and haveso few turns that they can be quickly wound byhand, whilst the. caracitors are small metallizedpaper ones, the maximum value being 0.01 ,II.F.

Constructional Aspects

For the experimental work on this system, the cir-cuit was built in the manner shown in the accom-panying photographs, no attempt being made toproduce a compact layout. All the components

WmELESS WORLD, NOVEMBER 1963

Two views of the microphone with its associated oscillator, phase-sensitive rectifrer and r.f. filter.

541

OSCILLATOR PHASE-SENSITIVERECTIFIER

BIFllAR r:1,:'We-UND L~l

,~MICROPHONEyt.

TOMICROPHONE

CIRCUIT........-----Z~~

,Fig.5. Microphoneamplifier.

employed can be of very small physical size, how-ever, thus permitting the final version to be builtinside a microphone casing t in diameter and 6 inlong. The smallest size of Mullard "red series"Vinkor is very gati!)factory for the oscillator coil.

D. C. Supply via Signal Cable:-In the early stagesof the work, the d.c. supply for the oscillator wasfed in along a separate pair of wires from thoseused for conveying the audio-frequency output-the wires may be seen in the photographs. Morerecently, however, the d.c. supply has been fed inalong the audio frequency cable, the necessaryarrangements for doing this being shown in Figs.4 and 5, for the microphone circuit and the micro-phone amplifier circuit respectively. The amplifieris that described 1n reference (3), suitably modifiedto apply the required d.c. voltage to the incomingline. The amplifier will give an output of 10 mVr.m.s. for any input between 0.15 mV r.m.s. andISO mV r.m.s., the harmonic distortion being under0.2 % throughout the whole of this 60 dB range.

With this scheme the only microphone cablerequired is an ordinary twisted and screened pair,such as might be used, say, with a moving-coilmicrophone-and it may be made up to at least 100yards long if required with no appreciable differencein performance.

It will also be seen, in the Fig. 4 circuit, that the

542

F1LHR

WIRE-WOUND

Fig. 4. Final microphonecircuit.

Hf220

2,200p

1:1

separate transformer originally used for feeding thereference voltage to the phase-sensitive rectifier hasbeen eliminated, an appropriate centre-tappedwinding being added to the oscillator transformer.

(To be continued)

REFERENCES

1. "Nanowatt Logic Using Field-effect Metal-oxideSemiconductor Triodes" by F. M. Wanlass & C. T.Sah (Fairchild). 1963 International Solid-State CircuitsConference Digest, p. 32.

2. "Non conductor Valves" by "Cathode Ray."Wireless World, March 1963, p. 145.

3. "Low-distortion Amplifiers (Part 2)" by P. J.Baxandall, Journal of the British Sound RecordingAssociation, Nov. 1961, p. 246.

It Wireless World Diary"THE answers to 1,001 technical and general questions'(from addresses of U.K. and overseas organizations totelevision standards and from u.h.f. television frequenciesto valve and transistor connections) will be found in the80-page reference section of the 1964 "Wireless WorldDiary". Now in its 46th year of publication the Diary-giving a week to an opening-is published by T. J. & ;J. Smith Ltd., and is available from newsagents and;booksellers or direct from this office. It costs 5s 6d in

I

irexine or 7s 6d in morocco leather, including purchase.tax. Overseas prices are 4s 8d and 6s 5d and postage .:is 4d. ,

WIRELESS WORLD, NOVEMBER 1963

12k+IZV

Ilk CARBON

3'3k

lOOp

OUT

33k l

I'

SJl.I "('

_lOOp<

S'6kI I

E