- Clarasonic : Distributor of ROHACELL
Transcription
- Clarasonic : Distributor of ROHACELL
COAXIAL AUDIO TRANSDUCR REVIEW AND CONCEPT DEVELOPMENT BY STEVE MOWRY The following discussion concerns the integration of a new electromagnetic induction drive tweeter and a midrange into a new coaxial audio transducer concept assembly. Four coaxial transducer implementations from high end loudspeakers will be reviewed before the new integrated coaxial transducer concept is presented. The primary design objectives were to minimize geometric diffraction potential and to achieve a high degree of manufacturability by utilizing modular assembly procedures. The target for rate of rejection for the final coaxial transducer assembly is no greater than 5% on the first manufacturing pass, while the rejected transducer assemblies can be subsequently reworked. Modular manufacturing streamlines incoming Quality Control and limits the supply chain while adding value at the part and subassembly source that is typically located in Asia. The induction drive tweeter diaphragm assembly was designed to be simple but robust with a two plane suspension system. The midrange was also designed to be simple yet robust and mechanically stable. The integration of the under-hung moving coil midrange and the EMI tweeter facilitate a motor assembly with a low reluctance AC flux path except for the midrange’s magnetic gap. Four 3-way Systems with Coaxial Midrange and Tweeter 1. TAD REFERNCE ONE: Retail price US$80,000 / pair (passive loudspeaker) with Be tweeter dome and Be midrange cone http://data.manualslib.com/pdf3/56/5548/554761pioneer/tadr1.pdf?48bb13604da8282d45ac88272a8e54c2 FIGURE 1. EXPLODED AND SECTIONAL ILLUSTRATIONS OF THE TAD CST COAXIAL TRANSDUCER Dual NdFeB magnet integrated motor assembly with serial magnetic gaps would seem to facilitate typical bottom-up gauged midrange assembly with quasi-butterfly tweeter diaphragm self-alignment. We will see later, that assembly can be simplified by utilizing modular assembly; however, this must be designed-in to the transducer assembly. FIGURE X. THE TAD R1 CST INTEGRATED COAXIAL TWEETER AND MIDRANGE WITH ISOLATION POT FIGURE X. THE TAD R1 LOUDSPEAKER FIGURE X. TAD R1 WOOFER 2 2. KEF BLADE: Retail price US$30,000 / pair (passive loudspeaker) with Al tweeter dome and Al mid cone www.my-hiend.com/leoyeh/2009c/White%20Paper.pdf FIGURE 3A. SECTIONAL ILLUSTRATION OF THE KEF BLADE MIDRANGE AND TWEETER COAXIAL TRANSDUCERS FIGURE 3B. SECTIONAL ILLUSTRATION OF THE KEF BLADE WOOFER FIGURE 4A. SECTIONAL ILLUSTRATION OF THE KEF BLADE TWEETER GP Acoustics filed a US Design Patent application for a Loudspeaker on 1 April 2011, www.freepatentsonline.com/D653237.pdf. FIGURE 4B. PICTURES OF THE KEF BLADE LOUDSPEAKER There is manufacturability issue related to gauging of the BLADE midrange voice coil. The midrange soft-part shim gauge must be pulled with only a single plane half-roll suspension attachment. The tweeter and the midrange are effectively individual 3 transducers mounted coaxially. The BLADE woofer uses 2 x spiders, which provides a two plane suspension attachment before the soft-part gauge is pulled. 3. THIEL 3.7CS: Retail price US$14,000 / pair (passive loudspeaker) with Al tweeter dome and Al mid diaphragm www.thielaudio.com/wp-content/docs/CS3.7-prem-tech-paper.pdf FIGURE 5. EXPLODED SECTIONAL AND ISO ILLUSTRATIONS OF THE THIEL 3.7CS MIDRAGE AND COAXIAL TRANSDUCERS There are manufacturability issues related to gauging. The soft-part shim gauge must be pulled with only a single plane suspension attachment, the spider. A more repeatable and robust topology would use 2 x spiders. See the BLADE woofer section illustration in figure 3B. The butt joint of voice coil bobbin to convoluted diaphragm could use a coupler to facilitate a more robust neck joint. The tweeter and midrange are effectively individual transducers. The tweeter is shown is figure 6. The assembly procedure for this transducer seems complicated and a bit mystical! FIGURE 6. ISO ILLUSTRATION OF THE THIEL TWEETER ON THE LEFT AND WITH THE FIVE MAGNET ARRAY ON THE RIGHT The 3.7CS tweeter illustrated in figure 6 uses five (5) NdFeB magnets, four (4) radially magnetized primary arc segment magnets and one (1) axially magnetize secondary ring magnet. 4 FIGURE 7. SECTIONAL ILLUSTRATION OF THE THIEL 3.7CS WOOFERS The THIEL 3.7CS woofer seems to be manufacture unfriendly. The soft-part gauge must be pulled with only the single spider serving to maintain the voice coil alignment while the cone-coupler and surround subassembly is attached. The topology is similar to a woofer that I did for Philips in 1999. See figure 9 below. The Al diaphragm can be treated as a dust cap and the cone is a coupler. SECTION VIEW FIGURE 8. PICTURE OF THE THIEL 3.7CS LOUDSPEAKER TOP VIEW Please note that the transducer illustrated in figure 9 had a unit cost of less than US$4.00 in 1999. FIGURE 9. CONE COUPLED Al DOMEDUST CAP WOOFER ASSEMBLY 5 4. GENELEC 8260A: Retail price US$14,000 / pair (active loudspeaker system with DSP hardware and software and with calibration microphone) with Al tweeter dome and Al mid cone www.genelec.com/documents/other/Genelec%208260A%20Technical%20Paper.pdf GENELEC OY filed a US Utility Patent application for a Nested Compound Loudspeaker Drive Unit on 19 July 2002, www.freepatentsonline.com/8660279.pdf . FIGURE 10. PICTURE OF THE GENELEC 8260A LOUDSPEAKER WITH A SECTIONAL DETAILS OF THE COAXIAL MIDRANGE AND TWEETER TRANSDUCERS IN AN INTEGRAL ISOLATION POT There are also manufacturability issues with the GENELEC 8260A. There is no spider(s). The soft-part assembly maybe manually buzzed-in where the assembly is moved around until no buzz, rub, or tic can be detected then the screws are tightened. Unfortunately, this is a random process and in the limit, the relative location of the voice coil is unknown. This is the only coaxial that uses quasi-butterfly self-alignment of the midrange voice coil without any shim gauging. Only shim gauging can guarantee a minimum clearance between the pole and the voice coil bobbin. Remember that “Quality” is defined as the absence of variability. The 19 mm button tweeter is consistent with a US$5.00 or less transducer in reasonable quantities. In general, when it is difficult to identify the transducer assembly procedure, then chances are it is difficult to assemble that transducer. 6 EMI TWEETER MODEL PRIMARY Ve Rp(i ) Lp(i ) j2f t Rem(i ) (1- I(t,i ) Lem(i ) I(t,i ) N :1 SECONDARY B2 rs(i ):1 Ls(i ) Rs(i ) . Blx(t,i ) 0.5BlI(t,i ) . x(t,i ) 1:Sd Rms(i) Mms Cms(i ) BlI(t,i ) NI(t,i ) FIGURE 11. NONLINEAR MODEL OF AN EMI TWEETER Where V is the AC input voltage amplitude at the transducer terminals (V). I is the input AC current amplitude (A). t is time (s) and f is frequency (Hz or 1/s). j is the unit-less complex operator, where j 1 . i is complex AC current (A). e is Euler’s Number. Re is the DC resistance of the voice coil (). Le is the AC self-inductance of the voice coil (H). Reb is the additional AC resistance related to eddy current (). Leb is the additional AC inductance related to eddy current (). RP is the DC resistance of the primary coil (). RS is the DC resistance of the secondary turn (). LP is the AC self-inductance of the primary (). LS is the AC self-inductance of the secondary (). Rem is the real part of induction loss (). Lem is the imaginary part of induction loss (). T is temperature (C) where 20C is the ambient. I is the induced AC current amplitude that couples to the secondary (A) Where is the loss factor, such that 0 < < 1.0. N is the number of turns in the primary coil. rs is the radius of the secondary (m). x is the position of the moving assembly (m). x is the change in position with respect to time, dx/dt or the velocity (m/s). B is the static DC magnetic flux density (T). l is the effective of length of the windings, where l = N2pr (m). . B(t) is the AC magnetic flux density (T). Rms is the mechanical loss term (s/m). Mms is the moving mass including air load (kg). Cms is the compliance of the suspension (mm/N). Sd is the effective piston area of the diaphragm (m2). p is the acoustic pressure (N/m2). A major design challenge for an EMI tweeter is the induction efficiency (a < 1.0) and the related sensitivity of the EMI tweeter implementation. 7 p(t,i ) FARADY’S, OHM’S, AND LENTZ’S LAWS Faraday was credited with the discovery of EMI in 1831. He found that the time, t (s), varying Electromotive Force (EMF), v(t) (V), produced around a closed path is proportional to the number of turns of coil wire, N, and the rate of change of the AC magnetic flux, (t) (Tm2), through any surface bounded by that path but in the opposite direction (), Lentz’s Law. d (t) v (t ) N (V) Faraday’s Law of Induction (1) dt d j 2 f (1/s) Assume (2) dt Then V j 2f N (V) (3) The impedance of a coil that is related to inductance is as follows. Z = j2 f Le Where Le is the self-inductance of the coil (H). V j 2π f N Where Le N and V = ZI Ohm’s Law. I I I VDROP j 2π f N VINDUCED j 2π f N Then in the limit, (4) (5,6) (7) (8) VDROP 1 Lentz’s Law. VINDUCED (9) EMI PATENTS Sony has spent much time on R&D and they have several patents that discuss and develop the theory for the above. www.freepatentsonline.com/5062140.pdf www.freepatentsonline.com/6904158.pdf www.freepatentsonline.com/7016515.pdf Klipsch also has two Euro Patents on EMI drivers. www.freepatentsonline.com/EP0453130A2.pdf www.freepatentsonline.com/EP0453130A3.pdf Then there is Tannoy’s co-axial. www.freepatentsonline.com/4965839.pdf I found a US Patent issued to W. K. Volkers in 1950. www.freepatentsonline.com/2494918.pdf I found another US Patent issued to S. E. Karlsson et al. in 1950. www.freepatentsonline.com/2621261.pdf Finally, the oldest Patent that I found was issued to A. Nyman in 1927. www.freepatentsonline.com/1643169.pdf Wisdom Audio has the latest EMI Patent that is based on Marshall Leach et al. R&D. www.freepatentsonline.com/US8009857.pdf 8 EMI PAPER www.redrockacoustics.com/induction_%20drive_%20loudspeakers_alma.pdf With regards to all the previous listed Patents, only US8009857 Bohlender et al. describes EMI motor assemblies that contain low reluctance AC flux paths. Sony, Klipsch, and Tannoy all utilize typical OD ring magnet motor topology (see figure 9 below); however, Tannoy’s transducer is an integrated coaxial where the low frequency driver’s voice coil is also the primary coil for the EMI high frequency transducer. Then in Tannoy’s case both the primary and secondary are suspended in the magnetic gap. Although interesting, the resultant acoustic output is not really hi-fi. Tannoy’s implementation seems more cost driven than performance driven. MOVING COIL DRIVES Cu SHORTING Al WIRE RING VOICE COIL Cu SHORTING RING Al WIRE VOICE COIL INDUCTION DRIVES Cu WIRE PRIMARY HIGH RELUCTANCE (OPEN CIRCUIT) ALUMINUM SECONDARY Cu WIRE PRIMARY ALUMINUM SECONDARY HIGH RELUCTANCE (OPEN CIRCUIT) (SONY & KLIPSCH TOPOLOGY) INTEGRATED HYBRID COAXIAL DRIVES FIGURE 12. SECTIONAL ILLUSTRATIONS OF THE TYPICAL MOVING COIL AND INDUCTION DRIVE TOPOLOGIES WITH AN INTEGRATED HYBRID IMPLEMENTATION The illustrations in figure 12 show three topologies. The induction drive section shows the primary coil is held stationary, while the secondary (shorting ring) is suspended in the magnetic gap. However, the moving coil drive section shows the shorting ring held stationary, while the voice coil is suspended in the magnetic gap. The hybrid coaxial topology shows an implementation of both moving coil and induction drives. 9 Equation 8 shows that the induced signal goes with frequency. This tells us two things. Shorting rings are typically used in high end tweeters. Inductive dive works best for high frequency transducers. In both cases, the load inductance is reduced or the load is less reactive with either a shorting ring or induction drive. SM Audio Coaxial Transducer Concept: Target retail price US$2,000 / pair with Be tweeter dome and sandwich composite cone SECTION VIEW BACK VIEW FRONT VIEW OD Ø180 mm NOMINAL MIDRANGE CONE-SURROUND ASSEMBLY VOICE COILSPIDERS ASSEMBLY WITH GAUGE TWEETER DIAPHRAGM ASSEMBLY EXPLODED SECTION VIEW HARD-PART ASSEMBLY BASKETTERMINALS ASSEMBLY FIGURE 13. ILLUSTRATIONS OF THE SM AUDIO INTEGRATED MIDRANGE AND TWEETER COAXIAL TRANSDUCER CONCEPT A listing of module subassembly specifications and illustrations follow. C.1 Basket materials / process / finish / terminals: Glass filled Nylon / injection mold / none / push type midrange terminals attached 10 SECTION VIEW BACK VIEW FRONT VIEW FIGURE 14. ILLUSTARTIONS OF THE BASKET-TERMINALS ASSEMBLY Glass filled Nylon was the preferred basket material for the following reasons. 1. Reduced transducer mass 2. Reduced basket cost 3. Reduced tooling cost 4. Reduced induction loss 5. Increased material damping C.2 Hard-part assembly materials / assembly process: The cup-pole and the gaps plate parts are machined Carpenter Chrome Core 13XP stainless steel or equivalent. The two secondary magnet keeper plates are machined and/or blanked AISI 1010 annealed steel or better with zinc plate. The primary tweeter coil is wound with round copper magnet wire on a Kapton HN film bobbin/integral height gauge and terminated with push type terminals. Nickel plated sintered neodymium iron boron magnets with ground surfaces are a must. The assembly is adhesive bonded together with an alignment gauge and/or fixture. Due to the complexity of this assembly, it should be purchased assembled and magnetized. 11 SECTION VIEW SHARED PRIMARY NdFeB MAGNET TWEETER SECONDARY NdFeB MAGNET AISI 1010 STEEL AISI 1010 STEEL FRONT VIEW CHROME CORE 13-XP LOW RELUCTANCE AC PATH CHROME CORE 13-XP MIDRANGE SECONDARY NdFeB MAGNET BACK VIEW FIGURE 15. ILLUSTRATIONS OF THE HARD-PART ASSEMBLY WITH THE TWEETER PRIMARY COIL ATTACHED AND TERMINATED MATERIALS MAXIMUM PERMEABILITY SATURATION FLUX DENSITY ELECTRICAL CONDUCTIVITY ANNEALED AISI 1010 2750m0 2.0 T 6.7 x 106 1/W/m CHROME CORE 13-XP 3200m0 1.7 T 1.3 x 106 1/W/m NdFeB MAGNET ~m0 NA 0.7 x 106 1/W/m TABLE 1. MATERIAL PROPERTIES OF HARD-PARTS C.3 Tweeter diaphragm materials / assembly process: Acoustic grade Beryllium foil dome with unalloyed aluminum secondary, acoustic grade poly-foam surround and Kapton JP film hot formed spider mounted on a black glass filled Nylon frame / adhesive bond with alignment gauge and fixture. SECTION FOAM KAPTON JP VIEW SURROUND SPIDER Be FOIL DOME FRONT VIEW GLASS FILLED NYLON FRAME Al 1199 SECONDARY BACK VIEW FIGURE 16. ILLUSTRATIONS OF THE DIAPHRAGM ASSEMBLY 12 C.4 Voice coil – Spiders materials / assembly process: Anodized aluminum foil bobbin with 15% copper clad aluminum round magnet wire coil and Nomex paper collar – hot formed and trimmed cloth blend spiders with molded nylon spacer / adhesive bond spiders to voice coil bobbin and spacer ring with alignment fixture. SECTION MIRROR CLOTH VIEW Al FOIL BOBBIN FRONT VIEW BLEND SPIDERS Cu CCAl TINSELS MAGNET ATTACHED WIRE & TINNED POLYMER RING BACK VIEW FIGURE 17. ILLUSTRATIONS OF THE MIDRANGE VOICE COILSPIDERS ASSEMBLY C.5 Midrange sandwich cone materials / process: Woven Kevlar fiber with epoxy skins on a Rohacell foam core with integral thermally sealed poly-foam surrounds and aluminum alloy foil cone to voice coil coupler / adhesive bond in heated mold and trim 13 SECTION VIEW 2xSOFT-FOAM SURROUNDS HARD-FOAM CORE Al FOIL COUPLER 2xKEVLAR FIBER-EPOXY SKINS FRONT VIEW BACK VIEW FIGURE 18. ILLUSTRATIONS OF THE SANDWICH COMPOSITE CONE-SURROUND ASSEMBLY Asian specialty transducer and/or part manufacturers can be identified and qualified to make the five (5) transducer subassembly modules. Then final assembly of the modules is convenient and can be made efficient and effective regardless of location, if and only if, the geometry dimensioning and tolerancing are correct and well documented. CLARASONIC in Thailand (www.clarasonic.com/) is presently working on the development of the cone-surround assembly illustrated in figure 18. INDUCTION DRIVE TWEETER DRIVER DESIGN CONSIDERATIONS 1. 2. 3. 4. 5. 6. 7. Low reluctance AC flux path Low moving mass with Be dome and 25mm ID Al secondary and no lead dress Large dome 35mm OD BETA booster secondary magnet Utilize low conductivity ferromagnetic material to reduce induction loss Eliminate typical electrically conductive materials where possible Inherent loading of the tweeter diaphragm by the midrange cone geometry et al. MOVING COIL DRIVE MIDRANGE DRIVER DESIGN CONSIDERATIONS 1. 2. 3. 4. 5. 6. A stable and repeatable bottom-up gauged assembly procedure Continuous geometric integration with tweeter and enclosure baffle BETA booster complementary secondary magnet Cu shorting ring to reduce inductance nonlinearity and tweeter crosstalk Under-hung voice coil topology Polymer basket 14 The resultant transducer concept is a no nonsense integrated midrange and tweeter coaxial implementation that is well suited for manufacture by a small knowledge based company. The intension is to implement loudspeakers utilizing one coaxial transducer and one or more KOMODO 9 EM woofers per side. SECTIONAL VIEW 85 cm NOMINAL FIGURE 19. SECTIONAL ILLUSTATION OF THE LOUDSPEAKER’S TRANSDUCER ARRAY WITH FRONT BAFFLE 15 The Author: Steve Mowry is in his 20th consecutive year of exclusively practicing audio transducer engineering that all began within the BOSE transducer research group (TRES) in mid 1995 after obtaining BS and MS degrees in Electrical Engineering with highest distinction from URI. He has worked with audio transducer and system development teams worldwide in diverse environments. With respect to transducer manufacturers and part suppliers, Steve is conveniently stationed in Southeast Asia. He is currently an independent transducer researcher and consultant to the loudspeaker industry with strong interests in Design for Manufacture, Supply Chain Design, Documentation Control, Concurrent Product Development, and Concurrent Project Management. Steve Mowry 18 September 2014 16 THE DEVELOPMENT OF A DINOSAUR INTO DRAGONS BY STEVE MOWRY The following is a discussion of the long term development of a high performance low frequency transducer topology and woofer implementation. The focus of the transducer R&D during the development period was multimagnet symmetrical drive assemblies as a way to improve low frequency transducer performance. Overall the developmental transducer topologies remained relatively consistent with incremental improvements with regards to the initial push-pull complementary symmetrical drive topology that was implemented in 2001. The high moving mass of low frequency transducers makes the typical cantilevered drive topologies a far cry from the ideal. The motivating consideration for the cantilevered drive topology seems to be at least in part convenience based. Two dimensional illustrations and linear parameter models were recreated from engineering notebooks and show examples of design milestones. Two high end manufactures’ low frequency transducer implementations are also illustrated with approximated dimensions. Common transducer topology elements can be observed within all of these illustrations. Additionally, the perceived pros and cons of each transducer implementation will be briefly discussed; however, the fundamental mechanical advantages of complementary symmetric topology should be at least somewhat obvious. In 2001 my employer asked me to design a new 6 in woofer that was intended to be a member of the “TM” (Top Most) product family. At that time, P.Audio was manufacturing and marketing their TM-12 woofer. The TM-12 was essentially reverse engineered by inspection of the Velodyne 12 (in) “Digital Drive” active subwoofer (http://velodyne.com/pdf/hgs/hgsx_series_datasheet.pdf) before I joined P.Audio’s transducer development team. A-TON Thailand, P.Audio’s sister company was the manufacturer of the Velodyne 12’s voice coil, at that time. Velodyne’s transducer implementation utilized common sized hard ferrite magnets with open tooling for much of the other parts. They used the popular NuWay large progressive roll spider with open tooling. The primary basket was the common “Venezuela” cast alloy basket, again with open tooling; however, the secondary basket required custom tooling. The additional parts were either selected and/or trimmed from open tooling parts or custom tooled. Both the Velodyne 12 and the TM-12 were of significantly different topology than anything that I had seen before. Finally there was a low frequency transducer topology that seemed to be “correct”. I used a high performance automotive analogy in code naming this topology as “Midengine”, with the “Wheelbase” defined as the peak to peak distance between suspension components. Being an electrical engineer by training, a complementary symmetrical configuration was something that I was familiar with and the analogy to modern electronics design was clear. I thought what would happen to their performance if amplifiers where designed asymmetrically like cantilevered drive audio transducers? The only complementary symmetrical drive transducers that I could identify then were either some planar magnetic or electrostatic panels. Their sonic characteristics are legendary but with limited low end output. Figure 1 contains an illustration of the TM-12 and a linear parameter model that were recreated from my engineering notebook. In all the following transducer illustrations (figures 1, 2, 3, 4, 5, 6, 9, and 11) the blue sections are the ferromagnetic low carbon steel parts and the cyan sections are the permanent magnets. Ø238mm 2xNuWay Progressive Spider 12in Venezuela Basket 2xMagnet OD 200mm ID 85mm Thick 20mm SECTIONAL VIEW A-TON Voice Coil TOP VIEW FIGURE 1. TM-12 aka RAPTOR 12 Transducer Assembly Drawing with a LspCAD Linear Parameter Model 2 Whereas the Velodyne 12 was used in a closed loop servo control system with an accelerometer on board the moving assembly, the TM-12 was marketed as a subwoofer driver without the accelerometer and/or control system. The TM-12’s parameters seemed to be suitable for car audio and there was consumer interest for that application. With respect to the LspCAD model in figure 1, it can be seen that mass and stiffness were high, while the ratio of BETA/mass/cone area was moderate, where b= (Bl)2/Re (N2/W). Using a single spider would reduce mass and stiffness. Plus using a smaller voice coil diameter would reduce mass and allow for a larger annular, less stiff, spider. With that information under consideration and with the typical cost, package size, and capability constraints, I designed an interesting small woofer. All parts except the terminals were new and required simulation, design, and documentation. HKS ABAQUS was used for soft-part design simulations, including spider, surround, cone, and dust cap, while VECTOR FIELDS OPERA was used for DC, AC, and Steady State Thermal motor design simulations. P.Audio’s management licensed that software for me in early 2000 to facilitate new audio transducer development in SE Asia. These were the same tools that were used at BOSE at that time. A transducer engineer is really only as good as his tools and his attitude. The primary customer for the TM-6 was MONACOR. They renamed this woofer “RAPTOR 6” (www.monacor.co.uk/products/carfisubwoofer/vnr/104350/) and this woofer is still sold today. An illustration of my transducer concept assembly from 2001 and a linear parameter LspCAD model that were recreated from my engineering notebook are shown in figure 2. 3 BOTTOM VIEW Ø120 VCID Ø26 2x23 2x8 30 17 SECTIONAL VIEW Ø118 TOP VIEW Dimensions are at nominal and in millimeters (mm) unless specified otherwise. FIGURE 2. TM-6 a.k.a. RAPTOR 6 Transducer Concept Assembly Drawing with a LspCAD Linear Parameter Model TM-6/RAPTOR 6 was well received by users especially in Europe. There are still a few videos on YouTube that contain demonstrations of the large displacement capability of this woofer. Here’s a link to an example, www.youtube.com/watch?v=9o32OsgSyPc. 4 A detailed listing of RAPTOR 6 part specifications follows. R.1 Gap plates material / process / finish: Low carbon annealed steel / stamp or blank / black zinc plate R.2 Magnets material / process / finish: Hard ferrite / sintered with ground surfaces / none R.3 Pole material / process / finish: Low carbon annealed steel / cold forge or extrude with secondary machining / black zinc plate R.4 Primary basket material / process / finish: Aluminum alloy / die cast with secondary machining / black paint R.5 Secondary basket material / process / finish: Aluminum alloy / die cast / black paint R.6 Voice coil assembly: 2 x 2-layer heavy build copper magnet wire coils wound on a polyimide film former with Nomex paper or equivalent collars and braided copper tinsels attached and tinned R.7 Spider material / process / color: Treated cloth / hot form and trim / tan R.8 Surround material / process: NBR with carbon black or equivalent / injection mold R.9 Cone material / process: Treated paper / wet formed, semi-pressed and trim R.10 Dust cap material / process: Matte coated paper / wet formed, hot pressed, and trim R.11 Terminals: Push type R.12 Fasteners: Blackened steel hex head or equivalent Sometime after leaving P.Audio to pursue independent transducer R&D, I developed a strong interest in multi-magnet motor assemblies including but not limited to the STEALLUS topology (STEALLUS, Part 1, Voice Coil, Nov. 2007 and STEALLUS, Part 2, Voice Coil, Jan. 2008), www.pearlhifi.com/06_Lit_Archive/14_Books_Tech_Papers/Mowry_Steve/Steallus_M otor_Design.pdf. Figure 3 contains a low frequency transducer implementation example illustration of the STEALLUS topology. 5 Ø223 VCID Ø64 4x12 43 2xQUASIINFINITE GAP 2x40 SECTIONAL VIEW Ø218 TOP VIEW Dimensions are at nominal and in millimeters (mm) unless specified otherwise. FIGURE 3. STEALLUS 11 Woofer Concept Assembly Drawing It should be clear that the RAPTOR 6 and the STEALLS 11 transducer topologies are related with the voice coils positioned between the surround and spider, mid-engine topology. STEALLS 11 began simply as an attempt to implement the “best” low frequency transducer possible. In retrospect, the resultant implementation could be considered an upgrade to the RAPTOR topology; however, this was not intentional. The fundamental difference between RAPTOR 6 and STEALLS 11 is related to the respective motor topologies, whereas RAPTOR 6 utilizes a push-pull topology, while STEALLUS 11 utilizes a simple stacked array of two magnetic return-less assemblies spaced by a nonferrous material. By using a spacer between the two magnetic assemblies, the resultant voice coil 6 configuration is Quasi-XBL2 (www.adireaudio.com/Files/XBL2TechPaper.pdf). The STEALLUS 11 is a simplified minimalist low frequency transducer implementation with very large and costly NdFeB magnets; however, performance potential is high for this topology. If the magnetic assembly is purchased as a single assembled unit and pre-magnetized, manufacture is just about as friendly as possible. A detailed listing of STEALLUS 11 part specifications follows. S.1 Gap plates material / process / finish: Low carbon annealed steel / stamp or blank / black zinc plate S.2 Magnets material / process / finish: Neodymium Iron Boron / sintered with ground surfaces / nickel plate S.3 Motor spacer material / process / finish: Nonferrous stainless steel / cold forge or extrude with secondary machining / none S.4 Primary basket material / process / finish: Aluminum alloy / die cast with secondary machining / black e-cote S.5 Voice coil assembly: 8-layer heavy build aluminum magnet wire coils wound on a polyimide film former with braided copper tinsels attached and tinned S.6 Spider material / process / color: Treated cloth blend / hot form and trim / tan S.7 Surround material / process: Polyether acoustic foam / hot form and trim S.8 Cone material / process: Carbon fiber-epoxy skins with Rohacell IG hard foam core or equivalent / join in mold with adhesive and trim S.9 Dust cap material / process: Carbon fiber-epoxy skins with Rohacell IG hard foam core or equivalent / join in mold with adhesive and trim S.10 Terminals: Push type S.11 Fastener: Blackened steel hex socket head or equivalent 7 Shortly after the STEALLUS discussion was published in Voice Coil, I noticed that other high end loudspeaker manufacturers were also developing interests in multi-magnet motor assemblies. Some of the top rated loudspeakers had also begun to utilize multi-magnet motors. Figure 4 contains an assembly drawing with estimated dimensions that illustrates the B&W 802D woofer’s topology, while figure 5 contains an assembly drawing with estimated dimensions that illustrates the KEF BLADE woofer’s topology. The transducers illustrated in figures 4 and 5 will serve as points of reference for the purpose of a relative transducer evaluation. The B&W 802D loudspeakers retail for US$15,000 / pair, while the KEF BLADE loudspeakers retail for US$30,000 / pair. These are not active loudspeakers; both systems are passive loudspeakers and require power amplifiers. The B&W 802D woofer has a topology that uses Neodymium Iron Boron magnets and resembles the STEALLUS topology but with a magnetic return path that is magnetically shorted at the very bottom of the motor, www.bowers-wilkins.com/Discover/Discover/Technologies/Dual-MagnetBass-Motor-System.html. However, that’s about all that is similar. This is obviously an asymmetric topology; the voice coil is overhung and cantilevered; and the voice coil bobbin is large and very long. The spider also seems undersized. There is a thick composite cone with a large dome dust cap. All in all, this woofer appears to be an attempt to take the typical low frequency transducer topology to a higher level of performance by using limited motor drive symmetry. 8 Ø200mm EMPTY SECONDARY GAP SECTIONAL VIEW SHORTED GAP TOP VIEW FIGURE 4. B&W 802D Woofer Assembly Drawing The KEF BLADE woofer shown in figure 5 uses opposed hard ferrite magnets about a gap plate in a configuration not too different from B&W 802D woofer; however the inside secondary magnet has ferrous material on only one surface, like a missing magnet keeper. This results in a quasiinfinite return with a very low magnet DC operating point for a magnet with moderate thickness and thus a less than robust motor implementation. Cold northern temperatures could cause some permanent demagnetization of that secondary magnet. There is again obvious asymmetric topology with an overhung large voice coil configuration. 9 Ø225mm QUASIINFINITE GAP VENTED COUPLER SECTIONAL VIEW SHORTED GAP ANALOGY TOP VIEW FIGURE 5. KEF BLADE Woofer Assembly Drawing With respect to figure 5, the KEF BLADE woofer’s soft-parts are quit extraordinary. The inverted shallow dome diaphragm will have limited geometric stiffness. So unless the diaphragm is a super material such Beryllium or Silicon Carbide, there is little hope for a wide band piston mode. Additionally the “Diving Board” diaphragm to surround termination is of concern. Diaphragms typically begin to bend at the OD as frequency is increased. Two spiders double the effective spider stiffness and require an additional adhesive neck joint. I believe that KEF uses two small spiders to stabilize the voice coil location during assembly so the soft-part shim gauge can be removed (pulled) before the cone-surround assembly is attached. This allows elimination of the dust cap but displacement will be constrained by the small spiders’ free length. The diaphragm is also the dust cap in this case. When I was a transducer engineer at BOSE, there was a rumor that an engineer was fired for implementing a reverse roll surround in a small woofer. This was only a rumor; however, reverse roll surrounds were simply banned at BOSE at that time. 10 KEF touts a “Vented Coupler” that couples the voice coil bobbin to the diaphragm. Their claims are narrated in a short video that can be view at YouTube, www.youtube.com/watch?v=CiAVX8_QWxI. Venting under the dust cap or in this case diaphragm seems like a good idea. Several manufactures implement similar venting with holes in the voice coil bobbin or the cone/diaphragm. Care must be taken because holes in the voice coil bobbin will encourage bending or in some cases buckling and/or yield. Finally and obviously the KEF BLADE woofer’s transducer topology is asymmetric and the voice coil is cantilevered. I just cannot understand the design objectives for such a woofer that is used in a US$30,000 loudspeaker system. Clearly my low frequency transducer design objectives are inconsistent with KEF’s. Having said that, the design objective for a new woofer concept was to maximize electromagnetic, mechanical, and acoustical performance with a hard ferrite magnet based motor assembly. Figure 6 contains the concept transducer assembly drawing of KOMODO 9 and a simulated linear parameter LspCAD model. All KOMODO 9 parts are unique except the spider and terminals. The spider geometry is equivalent to the TM6/RAPTOR 6 spider but it is flipped over to facilitate attachment, the bottom is now the top. It seems that I got the spider right back in 2001. 11 CENTER (PRIMARY) VOICE COIL BOTTOM (SECONDARY) VOICE COIL TOP (SECONDARY) VOICE COIL BOTTOM VIEW Ø180 23 2x25 25 4x4 2x12.5 VCID Ø26 SECTIONAL VIEW Ø118 TOP VIEW Dimensions are at nominal and in millimeters (mm) unless specified otherwise. Here’s Alan Babb’s US Utility Patent for an asymmetric dive cantilevered single voice coil implementation with multi-permanent magnets, Multiple Magnetic Air Gap Motor, filed 14 February 2011, www.freepatentsonline.com/8781150.pdf. FIGURE 6. KOMODO 9 Woofer Concept Assembly Drawing with a LspCAD Linear Parameter Model 12 KOMODO 9 is another attempt to implement the “best” low frequency transducer possible. Ironically, the resultant implementation can also be considered an upgrade to the RAPTOR 6; however, again this was unintentional. KOMODO 9 utilizes two of the push-pull type motors but in a stacked symmetrical array such that a primary magnetic gap is implemented along with two secondary magnetic gaps about that primary magnetic gap. Figure 7 contains a one half (right hand side) sectional contour plot of the DC flux distribution simulation for the KOMODO 9 hard-part assembly. The two magnets are magnetized in opposing directions. The primary and secondary magnetic gaps show the focus of DC flux density. Notice that the center pole carries only the DC flux from a single magnet in the stacked topology mirror array. This facilitates a small diameter voice coil. This is a critical design aspect and allows larger annular spider geometry. The bobbin must be long to facilitate the axial stack-up. A large diameter voice coil would thus be very high mass and would limit sensitivity. FIGURE 7. VF OPERA DC FEA Flux Contour Plot of the KOMODO 9 Permanent Magnetic Assembly Figure 8 contains a plot of a DC simulation of the nonlinear parameter Bl(x) for the KOMODO 9 motor assembly; where B(x) is the position dependent magnitude of the flux density (T); l is the length of the voice coil conductor and l = N2pr (m) where N is the number of voice coil winding turns and r is the radius of the voice coil. Then x is the position of the voice coil (m). 13 The data used to generate the Bl(x) plot shown is figure 8 was acquired using proprietary voice coil emulating command files that run within the Finite Element Model. The voice coil is segmented into three (3) segments and the motor has three (3) magnetic gaps. The voice coil could be considered to be a linear commutation; however, any switching must be performed electronically. The obvious control system methodology is DSP where each voice coil could have its own power amp. FIGURE 8. KOMODO 9 Large Signal DC Simulation of Bl(x) With regards to figure 8, the solid blue trace is the resultant DC magnetic flux linkage to the primary voice coil. While the dashed blue traces are the respective DC magnetic flux linkage to the two secondary voice coils. Then the solid black trace is just the sum of the three blue traces or the three coils driven with the same signal. The area under the black trace is then the work done per ampere. Wmax I Xmax -Xmax Bl(x) dx (Nm) Where I is the input AC current amplitude (A). What I call the “motor bias” is controlled by the distance between the primary and secondary voice coils. The bias for KOMODO 9 was selected for maximum linearity with minimum ripple. It should be noted that the solid blue trace related to the primary voice coil is typical for an overhung voice coil motor topology and is inherently nonlinear. It should also be noted that the roll-off slope of the black trace beyond the linear region (~ 12.5 mm) helps to control excessive displacement. This phenomenon is the result of the primary magnetic gap flux linkage plus the secondary anti14 magnetic gaps flux linkage. With regards to the primary voice coil (solid blue trace), when |x| > 17 mm, then Bl < 0! Then it can be observed that the resultant Bl(x) characteristic (the solid black trace) resembles what would be expected from an under hung bias. With all voice coil driven Bl(x) is linear and symmetrical about the rest position (x = 0) with just a bit of ripple. The objective is for 25 mm peak to peak displacement with reasonably low distortion at the lowest frequencies, 30 50 Hz. Several other bias options are convenient. For example, the magnetic assemblies could be spaced similar to STEALLUS 11 and an enhanced quasi-XBL2 bias or even enhanced under hung bias could be realized. However, the length of the voice coil bobbin and the overall axial stack-up would increase by the thickness of the spacer or additional gap plate thickness. KOMODO 9 is intended to be used in 3-way loudspeaker systems with midrange and tweeter transducers. A detailed listing of KOMODO 9 part specifications follows. K.1 Gap plates material / process / finish: Low carbon annealed steel / stamp or blank / black zinc plate K.2 Magnets material / process / finish: Hard ferrite / sintered with ground surfaces / none K.3 Pole material / process / finish: Low carbon annealed steel / cold forge or extrude with secondary machining / black zinc plate K.4 Basket material / process / finish: Aluminum alloy / die cast with secondary machining / black electro-plate K.5 Clamp ring / process / finish: Aluminum alloy / stamp or blank / black electro-plate K.6 Voice coil assembly: 3 x 2-layer heavy build copper clad aluminum magnet wire coils wound on an anodized aluminum foil former with Nomex paper or equivalent collars and insulated braided copper tinsels attached and tinned K.7 Spider material / process / color: Treated cloth blend / hot form and trim / don’t care K.8 Surround material / process: NBR with carbon black or equivalent / injection mold 15 K.9 Diaphragm material / process: Flat sandwich composite with carbon fiber-epoxy skins with Rohacell IG hard foam core or equivalent / join in mold with adhesive and trim K.10 Dust cap material / process: Matte coated paper / wet formed and hot pressed K.11 Terminals: Push type K.12 Fasteners: Blackened nonferrous countersunk hex socket head or equivalent K.13 Pole cap material / process / finish: Aluminum alloy / forge or extrude / black electro-plate Now in retrospect, the TM-6 was one of the “best” production woofers that I ever designed! The test of time serves as a testament to this claim. However, after more than a decade of independent transducer R&D, there is room for improvement and an upgrade could be interesting. My claim is then that KOMODO 9 is a serious upgrade to RAPTOR 6 with regards to almost all aspects of performance and sets a new standard for low frequency transducer topology. The KOMODO topology utilizes a flat composite diaphragm with coupling vents under the dust cap. Ironically, it was KEF that pioneered the flat diaphragm for low frequency transducers (www.kef.com/uploads/files/en/kef_units/A%20History%20of%20Kef%20 Drive%20Units%20issue%203.pdf). However, the flat panel composite technology has significantly advanced since the B139 woofer et al. and composite panels are now commonly utilized in the aerospace industry and in F1 auto racing applications. The flat diaphragm assembly significantly reduces the baffle cavity created by cone drivers. The basket and the surround termination are designed to facilitate flush mounting of the transducer from the backside of the baffle. The KEF BLADE woofer also seems to be designed to address the cone cavity et al. The KOMODO topology transducers utilize a trick molded surround that allows for high displacement while relatively increasing effective diaphragm area. The KEF BLADE and the B&W 802D woofers appear to use well dated half roll surround topology. I am presently working with Clarasonic in Thailand to develop flat composite diaphragms and attach trick surrounds for low frequency transducer applications (www.clarasonic.com/). The center of moving mass with the KOMODO topology is as close to ideal as currently possible. While the resultant inherent mechanical stability related to the “ultra wide wheel base” is second to no topology that I know 16 of. The backdoor lead-out eliminates through the cone lead dress. The KOMODO 9 transducer concept is designed for manufacture and assembly and is well suited for manufacture within Asia. Another almost coincidental feature of KOMODO/RAPTOR transducer topology is that DC Electromagnets (http://en.wikipedia.org/wiki/Electromagnet) can be conveniently utilized as a replacement for the hard ferrite permanent magnet assemblies. In car audio applications, a clean 12 (14) VDC source is available; otherwise a DC voltage source is required. My research has shown that a size requirement guideline for aluminum magnet wire field coils is approximately the same as for hard ferrite magnets for a given b(x) (N2/W) target. Aluminum magnet wire is the material of choice for two reasons, mass and cost. The mass density of hard ferrite magnet is approximately 4,850 kg/m3. Whereas, the mass density of copper is 7,760 kg/m3 and the mass density of aluminum is 2,700 kg/m3. So the aluminum magnet wire electromagnets can actually reduce transducer mass. Although, the cost of the aluminum magnet wire field coils will be significantly less than the equivalent copper magnet wire field coils they will still be significantly higher cost than hard ferrite magnets. In the cases with opposing magnet directions of magnetization such as KOMODO 9, the assembly procedure is simplified and obviously there is no need for a costly magnetizing system. The voice coil is shim gauged on both the inside diameter, ID, and the outside diameter, OD, during the assembly process to facilitate high quality and high reliability. Quality is defined as the absence of variability. Both the RAPTOR 6 and KOMODO 9 motor assemblies are lossy. The DC flux leakage about the magnet(s) OD is quite high. In contrast both the KOMODO EM 6 and KOMODO EM 9 are inherently magnetically shielded and the external to the magnetic assembly DC flux leakage will be low. This can be observed within the contour plots shown in figures 7 and 10. Figure 9 contains a concept assembly drawing of the KOMODO 9 EM low frequency transducer. I refer to KOMODO 9 EM as “DSP Ready”. There is the potential to actively control the electromagnets and the voice coils inputs independently and dynamically; however, that’s a problem for a digital signal processing specialist. 17 BOTTOM FIELD COIL TOP FIELD COIL BOTTOM VIEW SECTIONAL VIEW TOP VIEW FIGURE 9. KOMODO EM 9 Woofer Concept Assembly Drawing 18 FIGURE 10. VF OPERA DC FEA Flux Contour Plot of the KOMODO 9 EM Electro-magnetic Assembly Figure 10 contains a DC simulation contour plot of the magnetic flux. The DC magnetic flux goes as NI (ampere turns) from the field coils, where N is the number of magnet wire turns and I is the DC current. Figure 11 contains a concept assembly drawing of a TM-6/RAPTOR 6 that has been retrofitted with the KOMODO EM transducer topology with a 12 VDC electromagnet drive, an “Engine Swap” analogy. Two terminals for AC input signal are added to the bottom of the assembly, then the existing two terminals are utilized for the 12 VDC field coils supply. 19 BOTTOM VIEW Ø125mm SECTIONAL VIEW TOP VIEW FIGURE 11. KOMODO EM 6 Woofer Concept Assembly Drawing This discussion has briefly traced the R&D milestones that have taken low frequency transducer topology development from the Velodyne’s Digital Drive complementary symmetrical drive example to the KOMODO topology. Are there any serious claims that asymmetric transducer topologies with cantilevered voice coils represent an advantage in low frequency transducer implementations? I can only identify cost and/or inconvenience as the dominate obstacle to complementary symmetrical drive low frequency transducer implementations. 20 Finally, both STEALLUS and KOMODO transducer topologies resulted from independent R&D. Working independently allows for greater design flexibility, which can sometimes map into innovation; however, the team element of development is lost or constrained. Serious inquiries, comments, and/or questions are welcome and encouraged. They can be sent to Steve’s attention at [email protected]. The Author: Steve Mowry is in his 20th consecutive year of exclusively practicing audio transducer engineering that all began within the BOSE transducer research group (TRES) in mid 1995 after obtaining BS and MS degrees in Electrical Engineering with highest distinction from URI. He has worked with audio transducer and system development teams worldwide in diverse development environments. With respect to transducer manufacturers and part suppliers, Steve is conveniently stationed in Southeast Asia. He is currently an independent transducer researcher and consultant to the loudspeaker industry with strong interests in Design for Manufacture, Supply Chain Design, Concurrent Product Development, and Concurrent Project Management. Steve Mowry 18 September 2014 21