PAPERS On the Specification of Moving-Coil Drivers for Low-Frequency Horn-Loaded Loudspeakers * W. MARSHALL LEACH, JR. Georgia tnstute of Technology, School of Electrical Engineering, Atlanta, GA 30332, USA ‘A procedure is presented for the design from specifications of moving-coil divers for low-frequency horn-loaded loudspeakers. The method permits specification of the upper and lower system outof frequencies, the von ofthe cavity behind the driver, the driver aca, the ‘horn throat area, and the desired system electrical impedance. From these specifications, the required Thiele~ Small small-signal parameters and the electtomechanical parameters of the driver are determined under the condition of » maximum-sensitivity constraint on the system. ‘The procedure can be easily modified for a maximum-efficiency constraint, 0 INTRODUCTION Horn-foaded loudspeakers have been in widespread use for many years, especially in applications where large acoustic powers must be radiated and where control ofthe directivity pattern of the radiated sound is desired. Com- pared to the direct radiator, the high efficiency of a hom- loaded loudspeaker is probably its best known advantage. Forexample, a moving-coil driver that hasan electroacous- tic efficiency of less than 1% when used as piston radiator caneasily achieve an efficiency in the 10 S0% range when hom loaded. ‘When a given moving-coil driver is used in a hom sys- tem, the input impedance to the driver is inereased by the hom loading over the useful frequency band. This is be- cause the system compliance must be chosen so thatthe resonant frequency les atthe geometric mean between the lower and upper cutoff frequencies, Although the system total quality factor must be low for acceptable bandwidths, the in-band resonance of the horm-loaded driver cnn in. crease the input impedance suficiently so thatthe system sensitivity (that is, its acoustic output fora constant input voltage) becomes unacceptable. Thus an important consid- cration in a horn synthesis isthe system impedance, This rust be kept acceptably low so that adequate electrical input power can be obtained from moder amplifiers which are essentially constant-voltage sources with negligible ‘output impedance * Presented atthe 6Ist Convention of the Audio Engineering Society, New York, 1978 Nov. $= 6; revised 1979 September Thic paper investigates the system design aspects of low-frequency hom-loaded loudspeakers. First, as a back: ground, a complete electroacoustic analysis of hom-loaded systems is given, Based on this analysis, two synthesis procedures are then presented, one for a given driver and tone from specifications. The latter procedure allows the designer ta specify the Thiele— Small small-signal param: eters for the driver from the system specifications while at the same time constraining the system impedance to lie in an acceptable range for modem power amplifiers, To illustrate the synthesis procedures, two numerical examples are given. 1 GLOSSARY OF SYMBOLS B—— Magnetie lux density in driver air gap ¢ ‘Velocity of sound in air (345 m/s) Cay Acoustical compliance of air in box Cay Acoustical compliance of air in front chamber Caz Total acoustical compliance of driver and box Cro Mechanical compliance of driver suspension ee Open-circuit output voltage of electrical source fe Resonant frequency of driver on box fo Mechanical force on driver diaphragm Ji Lower midband cutot frequency of system fa Upper midband cutoff frequency of system Driver free-air resonant frequency ‘fo Hom cutoff frequency fs High-frequency band upper minus 3-dB fre- quency 90 JOURNAL OF THE AUDIO ENGINEERING SOCIETY, 1978 DECEMBER, VOLUME 27, NUMBER 12F(3)__ Midband system transfer function Fy(s) High-frequency band system transfer function ig, Voice-coil current supplied by electrical source I _Effective length of voice-cotl conductor in mag- netic gap Ly Inductance of driver voice coil Maa Acoustical mass of driver diaphragm assembly ‘Max Acoustical mass of horn throat impedance Mary Mechanical mass of driver diaphragm assembly Pax Acoustical power radiated Py. Blectrical input power to driver Ove Quality factor of system af considering elec- trical losses only Qes Quality factor of driver at, considering electrical tosses only Die Quality factor of eystom at fe considering acous- tical radiation losses only Que Quality factor of system at f. considering mechan- ical losses in driver and acoustical losses in box Qre Total quality factor of system atfe Rag Acoustical resistance of box losses caused by internal enerpy ahsorption Raz Acoustical resistance of driver electrical losses Ry. Acoustical resistance of hom throat impedance Raw Acoustical resistance of driver mechanical losses and box acoustical losses Ray Total acoustical resistance (Rax + Ran) Ry De resistance of driver voice coil plus output re sistance of source Res Electrical resistance of driver suspension tosses, bun acoustical losses, and acoustical radiation losses Ruy Mechanical resistance of driver suspension s Complex frequency (0 + jw) Sp Effective piston area of driver diaphragm Se Areaof hom throat uy Mechanical velocity of driver diaphragm Us Volume velocity emitted by driver diaphragm. Vas Volume of air having same acoustical compliance as driver suspension Vp Volume of box that loads rea of Utiver diaphragm Ve Volume of front chamber between driver dia- phragm and throat of hom Volume compliance ratio (Vs/Vs) 1 Powerefficiency ratio bo Density of air (1.18 kg/m’) prc Characteristic impedance of air (407 mke zayle) @, Angular resonant frequency of driver on box Lower midband angular cutoff frequency of sys- tem ‘oy Upper midband angular cutof? frequency of sys 1, Driver free.nir angular resonant frequency @, Hom angular cutoff frequency @; High-frequency band, upper minus 3-dB angular frequency 2 HORNELECTROACOUSTIC CIRCUIT ‘The point of departure is a basic introduction (or review, as the case may be) of the horn-loaded loudspeaker. With MOVING COM. DRIVERS FOR LOW-FREQUENCY LOUDSPEAKERS, two exceptions, the approach is based onthe electroacous tic circuit model of a moving-coil loudspeaker loaded by an acoustic hom as described in [1 and 2}. The frst exception 1s tat @ gyTAUOF Model Is UsEU for the voiee coil of the driver, This eliminates the confusing parllel-clement mo- bility and admittance circuits from the analysis. The second is tha the output impedance of the driving source is a- sumed to be negligible, for this is the case with modem power amplifies. In contrast, the early analyses of kor, loudspeakers driven from amplifiers with moderately high ‘output impedances calculated the efficiency as the ratio of acoustic output power to maximum power available from the source [1]. The maximum available source power is inversely proportional to the source output impedance Since moder amplifier exhibit a very low output impe- dance which is typically less than 0.1 0, the maximum available output power is a meaningless specification, for amplifiers are not designed to drive load impedances this, low. Thus an alternate definition of efficiency is used. To correspond withthe modem notation used in loudspeaker analyses, the small-signal parameters defined by Thiele [3} and by Small [4] for ditect-radiator loudspeakers and later applied to hom loudspeakers by Small [5] and Keele [6] will be used. Fig. 1 slows the basic configuration of a hor-londed smoving-coil loudspeaker driver. The driver is modeled as having piston areaSp. The rear ofthe driveris loaded into a box of volume Va. Its front is loaded into a chamber of volume P> which is coupled into the horn with a throat area Sp. The electro-mechano-acoustical equivalent circuit of the system is given in Fig. 2. ig. 2is divided imo three pans: electrical, mechanical, and acoustical. The electrical part shows the moving-coil driver connected to generator of voltage e, which supplies fa current ig. Ry and Ly afe the voive-cvil resistance and inductance, retpectively. A gyrator with impedance BI is ed to couple the voice-coil circuit to the mechanical circuit as shown in the Appendix, where Bis the magnetic flux density inthe ar ap and! isthe length ofthe voice-coil conduetor in the magnetic fed Inthe mechanical pat ofthe circuit, force is voltage and velocity is current. The current isthe velocity with which the driver cone moves. The elements My. Ryo, and Cyp are the mechanical mass, resistance, and compliance, re spectively, associated withthe driver Cone and its suspen: sion. A transformer with a tums ratio equal tothe piston area S, of the cone couples the mechanical circuit tothe acoustical circuit In the acoustical part ofthe circuit, pressure is voltage and volume vetoes 1s current. The current Up isthe vol une velocity emitted by the driver cone. In this circuit Rg and Cup are the acoustical resistance and compliance, re= spectively, ofthe box of volume V9, and Cg isthe acousti- cal compliance of the front chamber. The input impedence tothe hom is modeled as an acoustical mass M x, in parallel ‘with an acoustical resistanve Ray. Toran infinite exponen til horn, for example, this model is valid only above the cutoff frequency ofthe hom. Also, both My. and Ry, are functions of frequency. However, for frequencies greater than the hom cutoff frequency they approach constants. For JOURNAL OF THE AUDIO ENGINEERING SOCIETY, 1979 DECEMAFR, VOLUME 27, NUMBER 12 otthe exponential horn these asymptotic values are [1] Mu = SE w aye BE ® Where pyc is the characteristic impedance of air and a is the hom cutoff frequency. At low frequencies the exact ‘hom impedance expression must be used. For the infinite exponential horn tis is [1] aunea WI SF 13 3 where Ris given by Eq. (2). Itcan be shown that Eq. (2) ‘aves the correct hom acoustical resistance for all horn geometries provided the frequency is sufficiently higher than the cutoff frequency. @ It is convenient to divide Rar into two parts —one part associated with the electrical losses and the other part as- sociated with the driver mechanical losses plus box acousti- cea tosses. This will be dove by defining ae SR Rae a Ru The analysis of the circuit in Fig. 4 can be simplified by defining three frequency ranges [1]. This approach is valid if the upper system cutoff frequency is sufficiently higher than the lower eutoff frequency. This is the case if the system is to exhibit acceptable bandwidth. For low fre- quencies, C,y can be replaced by an open circuit and May by a shor circuit, For mid-band frequencies Ma, and Ce are replaced by open circuits. For high frequencies Cz is replaced by 2 short circuit and My, by an open circuit Rass + Rae @ eo Le ~ Fig. 1. Basic configuration ofa low-frequency hom-loaded loudspeaker. Fig. 2. Complete lectro-mechano-acowstcal equivalent circuit of hom-loaded system. The acoustical output of the hom modeled by the circuit in Pig. 2 is the calculate this, it is convenient totransform the electrical and mechanical pars of the circuit into the acoustical part. This is given in Fig. 3, where the familiarimpedance transforma: tion for a transformer and that for a gyrator given in the Appendix have been used. The voice-coil inductance Lis neglected in most low-frequency loudspeaker analyses and. will be neglected in the following. When this is done, the circuitof Fig. 3can be reduced to that given in Fig. 4, where Muy ower dissipated in the resistor Rai. Te May = ‘Si “ BE Raw Ce ea o Cop = Se CuwCan © Sa®Cun + Cx 982 3 THE MID-FREQUENCY RANGE Ie is the mid-sfequency range that is of most interest because itis in tis band that the system efficiency is defined. The equivalent circuit for this range is given in Fig, 5. A straightforward analysis ofthis reat shows that the power delivered tothe hor, that isthe power dssiated in Rats function of fequency i 1 Be? Ry DSRS Rar Ra Pan |FGe) F = Leet RaeRa ~ DR Rar Ror PMP ® where Rar. is defined by Eq. (7) and where F(s) isthe system transfer function, which is given by (0./Qre)8 FS) = sept e SF (We/Qrc)5 + OF (10) JOURNAL OF THE AUDIO ENGINEERING SOCIETY. 1979 DECEMBER, VOLUME 27, NUMBER 12oy sheuo Fas Cas Fig, 3. Complete acoustical equivalent circuit of hom-oaded syste ‘The parameters w, and Orr are defined as the system resonant frequency and the total quality factor Q, respec- tively. These are given by o= an VM Car Ore RFR 2) In a conventional design the poles of F(s) are real (or equivalently Q;x. = 0.5) if the system i to have an accepta- ble bandwidth. In this case F(s) can be written Co, + on)s MO Trays + om) o where oy < oy and e, = Vox aay Ore = Se as) ‘A typical plot of the magnitude of F(jw) as a function of frequency is shown in Fig. 6. The frequencies. and «y will be defined as the lower and upper minus 3-€B cutof frequencies, respectively, ofthe system. These are given by eyo gic = VI= 40a 6, on = 3G U + VI 40%] an Ore ‘The mid-band efficiency is defined as the ratio of acousti- cal power radiated to total electrical input power at w = a Because the system transfer function is unity for @ = @, the acoustical power radiated ie given by Eq. (Q) with the substitution |F Ge.) # = 1. The electrical input power must be calculated from the electrical equivalent circuit which is Fig, 4, Reduced citcuit of Fig. 3 neglecting the voice-oilindu- JOURNAL OF THE AUDIO ENGINEERING SOGIETY, 1979 DECEMBER, VOLUME 27, NUMBER 12 “obtained by reflecting all mechanical and acoustical circuit elements in Fig. 2 back into the electrical circuit. For simplicity this will be done only for w = a, Because this is the frequency at which the efficiency is defined. With this assumption and the preceding midband approximations, the electrical equivalent circuit is purely resistive, Iti given in Fig. 7. In this figure Rys is given by Br Roao * S5(Rxe + Rad Res _BEP/SE as) Where Ray is defined by By. (8). The elecuical input power is thus given by Lg Pee eRe L ¢ 2 Re Rar + Raw Rar Fax a9) It thus follows that the mid-band efficiency is given by - Rak : (Ray Ry) Raw RD o Because Ra: is inversely proportional tthe hor throat axea Sy. itis of interest to determine what value of Ry, maximizes the efficiency. By differentiating Eq. (20) with respect 10 Ry. and equating the derivative 10 zero, the te a om Fig. 5. Simpliied mid-frequency acoustical circuit Fig, 6. Asymptotic plot of magnitude of Eg. (11) for mid-band response of nom-loaded system. 959LEACH maximum efficiency occurs when Rox = VRaw Rae ‘The required throat area can be determined from Eq. (2) Quite often, however, it may not be desired ro design a horn systom for maximum efficiency. Instead, it may be desired to design the system for maximum sensitivity. that is, for ‘maximum acoustical output for a given electrical input voltage. By differentiating Eq. (9) with respect to Ry, and equating the derivative to zero, it follows that maximum sensitivity occurs when en Ra = Rar (22) ‘The values of Rq. for maximum efficiency and maximum, sensitivity are not the same because Ray, affects the system electrical input impedance and, consequently, the value of Input power Py for a constant input voltage. Because the valuc of Rq, is smaller for maximum efficiency than for ‘maximum sensitivity. it can be seen from Eq. (18) that the electrical input impedance is higher for the maximum ef- ficiency condition, Thus a higher amplifier output voltage is required to drive this imereased impedance if the same acoustical output is to be obtained as for the maximum sensitivity condition However, the amplifier output power required to drive this impedance is ess. ‘Three loss mechanisms can be identified in the midband acoustical circuit of Fig. 5. These are the power dissipated in the two parts of Rqy defined by Eqs. (7) and (8) aun the power dissipated in R,.. Thus three quality factors can be defined: 1 Qe = ORG (23) page 1 Bet PSR Ras + Ra, - opie 4) ec Raw Car \ Qi = Rae ° “These are the quality factors associated with the losses in Rye sRayy and R gu, respectively. The total quality factor for the circuit is given by Ore = 26) Rae Raw + RadCar With these definitions the efficiency expression of Eq. (20) can be rewritten: . en ‘The value of Ra, whic rnaximizes 1) can be written tai YP + (Ran = ‘The value of Ray which maximizes the sensitivity can be written L Boe IO 5canCar oo ‘The corresponding efficiency is 1 Ore "TO -On ep 4 THE LOW-FREQUENCY RANGE ‘As defined in Section 2, the low-frequency range is that frequency band below which M xp in Fig. 4 can be approxi inated by a short citeuit amd Cyy by an open eircuit, The low-frequency acoustical equivalent cireuit is given in Fig. 8. A generalized analysis of the low-frequency range is impossible because of the dependence of Mand Rx, on the hom geometry and on frequency. Therefore only the analysis for an infinite exponemtia! horn will be presented When the expression for the input impedance to an in- finite exponential hom given by Eg. (2) is used for the impedance of the parallel My. and R,, combination in Fig 8, it follows from a straightforwand analysis thatthe real power delivered to this impedance is given by (32 ? oC where «is the hom cutoff frequency. It can be seen that P nis zero for wo = wy. Because the horn input impedance is imaginary for « < a [1], no radiation occurs at or below the cutoff frequency, Klipsch [7] described a novel way to increase the acous- tic output of the hom for frequencies just above the horn cutoff frequency that consisted of choosing the total com Pliance Cx $0 that wR, Car = 1 in Eg. (32). IF tis is uu, the second term in parentheses in the denominator of Fig. 7. Mid:band electrical equivalent circuit at system resonant frequency, 54 JOURNAL OF THE AUDIO ENGINEERING SOCIETY, 1979 DECEMBER, VOLUNE 27, NUMBER 12PAPERS this equation is zero when « = «iy. By making the de- ‘nominator smaller for frequencies at and just above wo, the value of Pay is effectively increased. The condition that wR a.Car = 1 isequivalentto equating the mass reactance fof the hom throat atthe horn cuofF frequency to the reac- tance of the total system compliance at this frequency. This is a powerful technique for improving the low-frequency system efficiency. The technique was later refined by Plach and Williams (8), [9] who called it “reactance annulling.” 5 THE HIGH-FREQUENCY RANGE {As defined in Section 2, the high-frequency range is that frequency band above which Cyr in Fig can be approxi- tated by » shor circuit and M,, by an open circuit. The high-frequency equivalent circuit is given in Fig. 9. It follows from a straightforward analysis that the power delivered to Ras inthis circuit, thai, the acoustical power radiate, is piven by LBP Rw Pan 2 SERED Ra + Rao [Fuad F 33) where Fy(s) is given by Fuls) Ra Rae Re ‘The upper minus 3-dB frequency for this expression would correspond to oy of the mid-frequency range if Cye (This assumes that the impedance of Cy in Fig. 5 can be neglected at oy.) Proper choice of Cap in Eg. (34) can extend the high- frequency response of the hom system. Although the analysis may not be accurate if the voice coil inductance ‘cannot be neglected, the values of C,y can be determined from a specification of the quality factor or Q for the securnl-oider low-pass response of F(a). Let this be wit ten in the form Q = 1/VE. It then follows that the required value forC gp is May 4 Car a OueCax 1 Gre (Ver where Ore and Qye are defined by Eqs. (12) and (25), respectively. The value of Var is given by Var = PoC*Cars where pp is the density of airandc isthe velocity of sound in air, The corresponding upper minus 3-4B frequency is found from Lot (0 1 TOc ‘Ore ~ Oe ‘The value of k which maximizes «is a function of the ratio Ras/ Rex. For example, for the maximum sensitivity condition, Ry, = Rar and Que = 20re. Itfollows then that MyoCaas? + (GRE ar (Vilar Ra) + VER ar Ra ~ 4Ran Ta % Baye faa Be ye RE ( yea + MovING-OOI ORIVERS FOR LOW-FREGUENCY LOUDSPEAKERS the value of k which maximizes a for this condition is 4. Thus@ is 0.5, and this corresponds toa critically damped alignment for (3). The coresponding values for yr and ovate Car = 0.686 QbcCar Gy) 1.099 08) Ore 6 SYSTEM DESIGN WITH A GIVEN DRIVER In a system design for a given moving-coil driver @ specification of oy, and ay in addition tothe driver parame ters Is sumclent 10 determine all system parameters. Let/, be the driver free-air resonant frequency, Vag its volume compliance. and Qrs.Qus. atid rs its electrical, mechani- cal, and total quality factors, respectively. A design pro- cedure based on these parameters is straightforward and is summarized inthe following, 1) Caleulate the System Resonant Frequency we. Froma. specification of a, and ox. «2 can be obtained from Eq (14, Ga) Is +1 rn + Raw 2) Calculate the System Total Quality Factor Oye. This is given by Eq, (15) 3) Calculate the System Compliance Ratio a. If itis assumed thatthe change in mass loading on the driver is negligible when it is mounted into the hom system, the required system compliance ratio is given by 39) 4 38) TD Caleutate the Box Volume Vg. The effective volume of airin the back cavity behind the drivers given by Vas vy= 8. BS. fia (40) Bas) 4+ VE=OF. 69 5) Calculate the System Elecrical Quality Factor Qxc Under the assumption stated in step 3, Qc is given by JOURNAL OF THE AUDIO ENGINEERING SOCIETY, 1978 DECEURER, VOLUME 27. NUMBER 12 985Que = Va +1 Ors: (4y ©) Calculate the System Acoustical Load Quality Factor Q.e. fallow from the dafinitione of Ore, Qces Ome: and re that Qc is calculated from wg - fF -t- td De ~ Oe ~ Dee ~ Onc ‘To evaluate this, a value for Qyge must be known. For zero losses in the box, this is given by Our = Va + | Ous However, the box witl limit this to alower value because of losses in Ryo of Fig. 2. Small [8] gives the typical values for Que of 2w 5 for systems using filling material and 510 10 for unfilled systems. His discussion was with reference to closed-box direct-radiator systems which would normally employ a larger Vg than the horn systems. Thus a larger Value Of Qc would be expected for the smaller box hor: loaded system, A mote accurate method for determining Q,y. would be to measure it with the driver mounted on the box but not coupled to the hom. The measurement proce- dure is given in {10}, 7) Calculate the Horn Throat Area Sq. The required thom throat area can be calculated from Es. (2) and (25). 1 is given by eV rsQue lat Ne «42 Sy a) where cis the velocity of sound in alr. 8) Calculate the System Efficiency 2. The system f- ficiency can be calculated from Eq. (27). The value should be greater than the efficiency for a closed-box direct- radiator system given in [10]. Otherwise the design has no ‘merit over a closed:-box system with the same driver. 9) Calculate the Optimum Front Cavity Volume Var The feont cavity volume which optimizes of Eq. (36) should be determined. This isa function of the ratio of Ryu toRer given by Ore. ~ Die Cre “ 10) Design the Horn for Proper Reactance Annulling. a tine me ine Pow Fig. 8. Simplified low-frequency acoustical circuit Some at Fig. 9. Simplinednigh-trequency acoustical circuit, The details of acoustic horn design are given in [1] and (21 Reactance annulling can be achieved with an exponential hom by proper choice ofthe hom flare constant. In proper sign, reactance annuilling must occur ata frequency less than that is, the hor cutoff frequency must be less than 0 if, is to be the lower minus 3-dB frequency of the system, In the following a system design example is given 61 System Specifications f= He =0m 6.3 Computed System Parameters feo Se = 126.5 He, 2 Ore = 0.29, 3) a= 6.90. 9) Vq = 2.40 107 me. 5) Qee = 0.59. ©) Que = 0.60 (assume Que = 10), 7) Sp = 0.0297 m*. 8) 1 = 46.7%. 9) The maximum value tor «ws occurs tor k = 4.1 ‘The values for‘, and Vp are f= Gt = 416K Vip = 1.26 107% m?. 10) Reactance annulling cap be achieved by designing, the hom for a cutaff frequency of (a+ Src VS = 15.8 fa Because this frequency ie greater than a, recctance annul. ling cannot be used with this design. 7 SYSTEM DESIGN FROM SPECIFICATIONS, Ina system design from specifications, the parameters of the driver are obtained which will cause the system to meet the desired specifications. When designing a system from specifications, the most important considerations are its siae, bandwidth, and efficieney. In addition the input impo- dance is also considered to be a necessary specification because it should fall into the range of impedances that audio power amplifiers are designed to drive. If itis too high, for example, the amplifter may be forced into clipping 958 JOURNAL OF THE AUDIO ENGINEERING SOCIETY, 1978 DECEMBER, VOLUME 27, NUMBER 12ceptable acoustic outputs are obtained. Therefore procedure to be presented here allows the de- signer to constrain this parameter. It will be assumed that a maximum sensitivity design is desired rather than a maxi- ‘mum efficiency one. The design procedure for a maximum efficiency design is similar but involves slightly more com- plicated expressions. 1) Establish the System Specifications. The size of the system is affected by the volume Vj of the box behind the driver, the piston area Sp of the driver, and the horn throat areaS>. The performance of the system is determined by its, lowerand upper cutoff frequencies ay, and én, respectively ‘The proper interface to the amplifier is determined by the input resistance Rig at the system resonant frequency and the de voice-coil resistance Ry. The design procedure re- guires a specification of dese seveu para 2) Calculate the System Resonant Frequency a. From a specification of oy and wy, @. can be obtained from Eq. ad), 3) Calculate the System Total Quality Factor Qxe. This is given by Eq. (15). 4) Calculate the System Electrical Quality Factor Qxe For the maximum sensitivity design, Qre = 20re- Ore is thus calculated from 1 1 1 Dec ~ Wee Ou “ The value that is used for Ouc is dependent on the box losses. Ths was discussed in step 6 of the preceding design procedure 3) Calculate the Total Stem Yotume Compliance Vx. ‘The roquired sytem volume compliance i stained fom Eq. (25) with the substitution Rar = Poc/ Sr, Qtc = 20rc, and Cyr = Var/byc® The results 2Orewe 6) Calculate he System Compliance Ratio «., The com- pliance ratio is given by Var 48) a Ve — Vax 1) Calculate the Driver Free-Air Electrical Quality Fac- tor Ops. Eq. (41) can be used to calculate Ors. Itis assumed thatthe assumption stated in step 3 of the preceding synthe- sis procedure holds, 8) Cateulure the Driver Free-dir Resonant Frequency ,. If the driver electrical losses predominate over the ‘mechanical losses, oy is approximately given by a- a) Vari ‘The conditions under which this relation holds arc dis: ceussed in [10], 9) Calculate the Driver Free-Air Volume Compliance Vs. Eq. (40) may be solved for 4s from the specified Vy and calculated a. 10) Calewlate the Driver Free-Air Mechanical Com pliance Cy. The required driver mechanical compliance is “obtained from its volume compliance by the relation [10] 'MOVING-COIL DRIVERS FOR LOW-FREQUENCY LOUDSPEAKERS Vine Tee ” Cow = 11) Cutcutue the Driver Moving Mechanical Mass ‘My. The total required moving mass is given by (10] 1 “ECan” Muy = oo 12) Catculate the Driver Electrical Resistance Res Due to Suspension Losses. If isthe desired input resistance at 1, and Re isthe de voice-coil resistance, Res is given by Rus = Rin on 13) Calculate the Driver Bl Product. The BI oF motor product ofthe driver is obtained from Eqs. (18), (23), (24), 2nd (20). The relation is Bl Ly pot Oe) ar 6) SBRes (Ga 14) Calculate the System Efficiency 7. This is given by Eq. G0). 15) Calculate the Volume Vay of the Front Cavity. From Eq. (37) the value of Vay for the maximum sensitivity design is Vag = 0.686 04 Far 63) The new upper minus 3-4B frequency is given by Eq. (38) 16) Design the Horn for Proper Annulling. This ts dis- cused instep 10 of the preceding synthesis procedure. Tn the following a system design example is given: 7.4 System Specifications = 0.015 m* 0.073 m* 0.018 m? 40 He. = 400 Hz 80 = 40 7.2 Computed Driver and System Parameters. Die 20.5 He. i 3) Ore = 0.29 4) Ore = 0.62 assume Qye = 10) 3) Var — 0.0135 mt 6a =88l 7) Ors = 0.20. s Se = 40.4 He, 9) Vas = 0.1320 10) Cup = 176% 10-4 /N 11) Myo = 0.0882 kg, 12) Ry = 40. 13) BE = 22.61-m 1) 1 = 43.9%. 18) Vay = 7.79 10-+ mt fy = 079m. JOURNAL OF THE AUDIO ENGINEERING SOCIETY, 1978 OCCEMBCN, VOLUME 27, NUMBER 12 osueAc 16) For proper reactance annulling. the hor cutoff fre- quency must be WDSc fa Ty ~ 15 Ne Because this is higher than f,, proper reactance annulling ‘cannot be achieved without a change in the system speci- fications. 8 CONCLUSIONS ‘The basic electroacoustic theory of the hom loudspeaker system fas been reviewed. Expressions for the acoustic ‘output from the system have been developed for the low, ‘mid-, and high-frequency bands. Two system synthesis procedures have been presented, one based on adesign with ‘given driver and the other based on system specifications, Numerical examples have been presented to illustrate ap- plication of the eynthesis procedures. 9 APPENDIX ‘THE GYRATOR MODEL OF A MOVING-COIL LOUDSPEAKER DRIVER The gyrator is a circuit element which makes a conve- nient model for the moving-coil loudspeaker voice coil, ‘The advantage of this model over the standard electroacous- tic transformer model Is thatthe confusing parallel-clement ‘admittance and mobility circuits ean be circumvented. Be- ‘cause the author had seen no applications in the literature of gyrators to loudspeaker analyses, this appendix is included. ‘The gyrator is a passive and lossless circuit element that is somewhat like the ideal transformer, except tha it inter- changes the roles of current and voltage at its output termi- ral pair. The basic circuit symbol for a gyrator is given in Fig. 10. The equations which define its terminal behavior are Ri (4) Y= Rh, oo where R is called the gyrator impedance. If the gyrator is terminated in an impedance Z,,, then v; = i,Z,. Simultane- ‘ous solution of this equation with Eqs. (34) and (55) yields the input impedance to the loaded gyrator: “LR Zz (36) En = ‘Thus the input impedance is the constant R* times the 1cviprocal ofthe load impedance Fig 10. Citenit symbol of a gyratnr used asa two-terminal-pait network, If the gyrator is driven at its input terminals from a ‘generator with open-circuit voltage v, and output impe- dance Z,, then, = ¥ —1,Z,. Simultaneous solution of this ‘equation with Eqs. (94) ana (93) ylelds me relationship between vy and Fn Fis 67 “This is equivalent to a voltage source of value (R/Z,)¥, in seties withthe impedance R?/Z,. The equivalent input and ‘output circuits for these two cases are given in Fig. 11. The basic electroacoustic equations which govern the ‘moving-coil loudspeaker involve the force fo generated on the voice coil when a current £ flows in it and the back clcctromotive force e generated when the voice coil moves. ‘The equations are (1): fo=Bli = Bluy (38) (59) where B is the magnetic field flux density in the air gap, is the length of voiee-coil wire in the magnetic Feld, and up is the velocity with which the voice coil moves. The gyrator ccan be used to represent these equations as shown in Fig. 12, In the more familiar transformer representation of these equations the force is the current and the velocity is the voltage atthe transformer secondary. This requires the use of parallel-element mobility-type mechanical equivalent circuits. In contrast the gyrator permits forming the more familiar series-element impedance type circuits directly ‘without the need w fora ciscuit duals, @ 2 2 Ea ) 1, Bquivalent input and output circuits of the gyrator. (a) [Equivalent input circuits (b) Equivalent output circuit. Fig. 12. Gyrator model for the electromagnet transducer of & moving-coil loudspeaker driver. ‘mechanical eo JOURNAL OF THE AUDIO ENGINEERING SOCIETY, 1979 DECEMBER. VOLUME 27, NUMBER 12PAPERS 10 REFERENCES: [1] L. L. Beranek, Acoustics. (McGraw-Hill, New York, 1954), ch. 9. [2] H. F. Olson, Elements of Acoustica Engineering. (Van Nostrand, New York, 1947), ch. 7 [3]. A. N. Thiele, “Loudspeakers in Vented Boxes, arts land Ii," J. Audio Eng. Soc., vol. 19, pp. 382~391 (1971 Mayy; pp. 471-483 (1971 June). [4] R.H_ Small. “Direct-Radiator Loudspeaker System Analysis,” J. Audio Eng. Soc., vol. 20, pp. 383-395 (1972 June). {5] RH, Small, “Suitability of Low-Froquoncy Drivers for Hom-Loaded Loudspeaker Systems.” presented at the 57th Convention of the Audio Engineering Society, Los ‘Angeles, 1977 May, preprint 1251 [6] D. B. Keele, “"Low-Frequency Hom Design Using MOVING-COIL ORIVERS FOR LOW.FREGUENGY LOUDSPEAKERE Thiele/Small Driver Parameters."” presented at the S7th Convention of the Audio Engineering Society, Los Angeles, 1977 May, preprint 1250. (7) PSY. Klipsch, “A Low-Frequency Hom of Small Dimensions." J. Acoust. Soc. Am., vol. 13, pp. 137= 144 (1941 Oct.) [8] D. J, Plach, "Design Factors in Hom-Lype Speak: ers,"" J. Audio Eng. Soc., vol. 1, pp. 276—281 (1983, Oct.) {9] D. J. Plach and P. B. Williams, “*Reactance Annu ling tor Hom-Iype Loudspeakers,” Kadio-Electron. Eng. pp. 1518 (1935 Feb.) [10] R. H, Small, “Closed-Box Loudspeaker Systems, art I: Analysis,” J. Audio Eng. Soe. vol. 20, pp. 798: 308 (1972 Dec. }; “Part M: Synthesis," vol. 21, pp. 11~ 18 (1973 Jan./Fob.) ‘THE AUTHOR \W. Marshall Leach, Jr.,received B.S. and M.S. degrees in electical engineering’ from the Univesity of South Carols, Colaba, 1962 ad 1964, and a Ph.D. degree in electrical engineering from the Georgia Institute of Technology, Atlanta in 1972 Ti 1968 worked wt NASA iu Hannpton, Vigna. From 1965 to 1968 he served as an officer in the U.S. Air Force Since 1972 he has been a faculty member at the Georgia Institute of Technology where he is presently associate professor of electrical engineering. His interests are applied clectromagnetics, audio and electroacoustes, and elec tronic circuit design Dr. Leach isa member ofthe IEEE. the Audio Engineer ing Society, Sigma Xi, Tau Beta Pi, Eta Kappa Nu, Omi ‘cron Delta Kappa, and Phi Beta Kappa.LETTERS TO THE EDITOR COMMENTS ON REACTANCE ANNULLING IN HORN-LOADED LOUDSPEAKER SYSTEMS Ina recent paper.' Leach mentions that the method of using system compliance reactance to cance! horn throat ‘mass reactance in the region above horn flare cutoffis “a powerful technique.” I could not agree more since thisis, pechaps the only case where the laws of physics aid the designer rather than frustrating him. However. Leach credits Klipsch? as the originator of the idea, and while his work was no doubt independent, Wenteand Thuras? used the same technique in the low-frequency horns for the Phitadetphia-Washington “Audio Perspective demonstration in the early 19306 In the interest of historical accuracy I believe that it was actually the late Al Thuras who first became aware of the potential of the method.* His patent® contains a complete and lucid description in the specification. Iti strange, however, that the technique was completely jenored in the claims of his patent. As Bell Labs. was wont to do, perhaps there isa separate patent covering this particular innovation, Epwanp F. McCain, Jk Morningside, MD 20746 Author's Reply I greatly appreciate Mr. MeClain’s comments on my paper and his historical perspective on the use of reac- tance annulling in horn-loaded loudspeaker systems. ‘When I wrote the paper I had intended to cover only the specification of the driver smal ignal parameters in a LW. M, Leach, Jr “On the Specification of Moving: Drivers for Low-Frequency Horn-Loaded Loudspeakers, Audio Eng. Soc. vol. 27. pp. 950-959 (1979 Dec.) Wrikiipech, A Lau. Froquoncy Harn of Small Diesen J. Acoust. Soc. Ant Nol: 13. pp. 137-148 (1941 Oct reprinted /. Audio Eng. Soc vol. 27, pp. 161-148 1979 Mar). "'E. C. Wente and A. L. Thuras, "Auditory Perspective— Loudspeakers and Microphones,” Trans. AUPE. vol. 53, pp 17-24) 934 Jan) reprinied J. dudio Eng. Soc, val. 26. Pp, 518-525 (1978 July/Aug.). (See first paragraph, right-hand column, page $21.) “In accordance with still another feature ofthis invention, the chamber adjacent one surlace of the diaphragm is propor™ toned ro that the stflpess thereof is substantially equal in ‘magnitude to the mean reactance component ofthe horn load Impedance throughout a range of audio requencles, whereby the reactance component of the horn load impedance is sub- Stantially neutralized throughout this range of frequencies (irom Thora) SA. L. Thuras, U.S. Patent 2,037,185, filed March 28, 1933, sssued April 14, 1936, Note that this patent had not been sued when the Wente and Thuras paper was published, ono4-ros4/o1a70se0-08400.75 system design and no details that pertain to acoustic horn design. However, because reactance annulling is sucha simple concept did include brief description of ivand its motivation, When I attempted 10 generate wo design examples to illustrate the synthesis procedures, | was unable to incorporate the concept as I understood it into my examples. That is, I could not force reactance annulling to occur at or below the desired lower mid- band cutolt frequency. 1 was convinced that the laws of physics were working against me ‘After the paper was published. I made a second at- tempt to incorporate reactance annulling at orbelow the lower midband cutoff frequency into the synthesis pro- cedures. However, | met with no success. Indeed, | found it straightforward to show that itis impossible to do with an exponential horn. T came to the conclusion that the lower midband cutoff frequency, which I denot- ed by f, must be specified artificially low in order to obtaina reactance annulling frequency that occursat or bbclow the desired lower cute hequency for the system. Although this modified procedure could be used for an acceptable design, the artificially low choice for f, would result in a reduced efficiency. In addition, it re= sults in an electroacoustic driver system which is capable of a wider bandwidth than the horn to which it is coupled. Fortunately this problem can be eliminated by choice of a hyperbolic horn*” rather than an exponen- tial horn. The following argument follows Plach’ and Salmon.* Salmon’s patent states, “Where the cut-off frequency of the horn (the frequency below which the acoustic resistance of the horn is zero) is below the resonant frequency of the driving motor with the impe- dance on the hora side of the diaphragm reduced t0 zero, itis possible to substantially match and annul the reactance of the driver, Which, inthis range, is capacita- tive, by means ofa properly selected member of the horn family herein described.” The parameter M in the fol- lowing identical to Plach's and Salmon’s parameter T- ‘The general formula for the cross-sectional area of the hyperbolic family of horns is given by 5 = 5, (somn(2) + arsins (2) where Sis the cross-sectional area af the throat, risthe axial distance from the throat, x, is the reference axial a feb Branek, Acoustics MeGraw-Hil, New York, 1958), 1D. J. Plach, “Design Factors in Horn-Type Speakers." J Audio Eng. Soe. vol. 1, pp. 276-281 (1953 Oct.) and Loud Speakers Audio Engineering Society, New York, 1980), SV. Salmon, U.S. Patent 2,338,262, filed 1942 July 23distance, and M is a parameter such that 0 1.0, the reactive values are inductive ACKNOWLEDGMENT ‘The author would like to gratefully acknowledge the assistance of Mr. Richard F. Long of Richard Long and ‘Associates, New York, in finding the errors in the author's paper and for suggesting the use of the hyper- bolic family of horns to achieve reactance annulling. He ‘would also like to acknowledge the aid of the reviewers in providing additional background material on reac tance annulling. ‘COMMENTS ON THE HISTORY OF MAGNETIC RECORDING In all the accounts of the history of magnetic record- ing that I have come across until now (e.g...the excellent and personal accounts of John T.Mullin), the devel- ‘opment is being described as having been continuous and universal until World War II, when high-frequency bias was developed and kept secret in Germany. It is said that high-frequency bias had not been rediscovered and made commercially feasible until 1945, when analy-