THOR: A D’Appolito Transmission Linemadisound.com/pdf/Kits/audioXpress Thor Review.pdfVoice-coil diameter 39mm Suspension compliance 1.6mm/N Voice-coil height 16mm Suspension resistance

During CES 2000 I met withNorwegian and US representa-tives of SEAS. They describedtheir new Excel line of drivers

to me and asked whether I could designa flagship loudspeaker using these driv-ers that would highlight their extraordi-nary capabilities. They wanted some-thing other than a vanilla sealed orvented box.

I suggested a transmission line,pointing out that the non-resonant be-havior of this enclosure assured that wewould hear the full capabilities of thesedrivers free of “boxy” coloration. TheSEAS folks agreed enthusiastically,and the THOR transmission-line projectwas born.

THE THOR-EXCEL TRANSMISSION-LINE LOUDSPEAKERTransmission-line (TL) loudspeakershave long enjoyed a small but dedicat-ed following, especially in the DIY com-munity. The advantages of TLs are wellknown. They are essentially non-reso-nant enclosures, producing a deep,well-controlled bass response. For agiven driver, bass response can oftenextend well below that produced witheither a vented or sealed enclosureusing the same driver. Above a few hun-dred Hz, the line-filling material com-pletely absorbs the driver back wave,giving the TL an open, non-boxy sound.

Unfortunately, the TL has not en-joyed wide commercial popularity dueto the lack of a good design theory andthe additional complexity of enclosurefabrication relative to the more conven-tional vented and sealed enclosures.Recently, however, work by G.L.Augspurger has appeared in the techni-

cal literature1 and in audioXpress2,3,4

that, while not providing a completetheory of design, has given us an excel-lent starting point. This, coupled withmodern PC-based acoustic measure-ment systems, allowed me to convergequickly to an optimum design for thenew Excel W18EX001 woofers.

The present design uses an MTMdriver configuration in a tapered, fold-ed line uniformly filled with Dacron pil-low stuffing. Tapering the line greatlyincreases the frequency range of bassaugmentation produced by the line.Using two mid-bass drivers exciting theline at slightly different points reducesmid-bass ripple.

The resulting line produces a uniform3–4dB bass response lift from 110Hz allthe way down to 20Hz with less than1dB ripple. The −3dB point is 44Hz. Con-trast this against 65Hz for a similarly

damped sealed enclosure. Below 45HzTL bass response falls off at 12dB per oc-tave, compared to the 24dB/octave fall-off rate of a vented system. In mostrooms useful bass response extendswell down into the 30Hz region.

Above 2500Hz the system crossesover to an Excel T25CF002 tweeter. Sev-eral hundred hours of laboratory test-ing and listening have gone into pro-ducing a seamless transition betweenthe mid-bass and tweeter drivers. Youliterally cannot tell where the woofersleave off and the tweeter begins.

The Excel product line from SEASwas introduced in 1994 as a showcasefor the company’s best ideas and tech-nologies. Originally comprised of onlyfive models, the Excel line has expand-ed to ten products, with additional de-signs in continuous development.

THE W18E001 WOOFERBuilding a “better” mid/woofer re-quired a complete rethinking of nearlyevery component in the driver: thecone, the magnet system, the surround,and the basket.

With this exceptional design for some exceptional drivers by an excep-

tional designer, transmission-line ownership is well within your grasp.

By Joe D’Appolito

THOR: A D’Appolito Transmission Line

8 audioXpress 5/02 www.audioXpress.com

PHOTO 1: In the lab, ready to test and trim the line.

The Magnesium ConeThe advantages of metal cones are wellknown. They remain virtually pistonicthroughout their passband, and do notsuffer from midband cone edge reso-nance problems so common in paperand other soft cones. Prior to the devel-opment of the magnesium cone, virtual-ly all metal cones used some form ofaluminum alloy. While aluminum is aneasy material to form either by stamp-ing or spinning, it also suffers from itsshare of acoustic drawbacks.

To keep the moving mass reasonablylow, the cone must be quite thin. For an18cm woofer, the nominal thickness isapproximately .18mm. This, unfortu-nately, results in a cone with numeroushigh Q breakup modes starting at about5kHz and extending beyond 10kHz.

SEAS therefore decided to search fora material with a potential for greaterstiffness than aluminum. Magnesiumwas attractive because its specific gravi-ty was only 1.7 versus 2.7 for aluminum.This meant that, for the same conemass, a magnesium diaphragm couldcontain almost 60% more material byvolume than aluminum. This gave thepotential for much greater stiffness andinternal damping of the cone with noincrease in mass. Acoustic testing ofprototype magnesium cones immedi-ately revealed the benefits over the alu-minum cone: the breakup modes hadbeen largely reduced to a single, well-defined peak that could easily be sup-pressed via simple notch filtering.

The question was how to produce thecone? Magnesium does not lend itselfto bending or shaping in the thicknessrequired for a loudspeaker cone. Thatleft the only option of die-casting. Fortu-nately, a small magnesium foundry

close to the factory was able to cast therough cone. But getting to the finishedcone would require that the remainingprocesses be developed in-house.

SEAS developed a special machin-ing process to remove the preciseamount of material necessary to shapethe cone and achieve the proper mass.Through much experimentation, acone of varying thickness between.26mm and .33mm was found to be theideal solution.

All that remained was the finishingprocess to give the cone an attractivelook and prevent it from corroding overtime. For this, a chemical etchingprocess was developed, followed by acoat of protective lacquer on the frontand rear surface, giving the cone itsunique appearance.

The Excel MotorTo gain the greatest advantage from themagnesium cone, an exceptional mag-net system was required. The key designgoals of the Excel motor were to: 1) Re-duce the levels of eddy current distor-tion and flux modulation, thereby reduc-ing harmonic and intermodulation dis-tortion, 2) Stabilize the inductance of thevoice coil under all excursion conditionsto reduce modulation distortion, and 3)Improve the heat transfer from the coiland pole piece to the outside air to re-duce voice-coil temperature and subse-quent voltage sensitivity modulation.

These goals were accomplished byincorporating two heavy copper ringsfitted above and below the magnet gapdefined by a T-shaped pole piece, whichwas press-fit into a bumped back plate.To further enhance the heat transfer ca-pability, a solid copper phase plug wasfitted to the top of the upper ring. The

stationary phase plug replaces a con-ventional dust cap and thereby elimi-nates the acoustic resonator behind thedust cap. At the same time, the excel-lent thermal conductivity of the phaseplug aids tremendously in heat dissipa-tion, while the air movement from thecone over the phase plug also serves tocool the motor.

The Excel BasketA high-performance motor and coneshould not be mechanically or acousti-cally limited by a less than optimal bas-ket. For the W18E001, an entirely newstate-of-the-art, die-cast zinc basket wasdeveloped. The casting is extremelystiff, maintaining precise alignment ofall mechanical parts, and providing astable and secure mounting surface forthe cabinet.

At the same time the rear of the bas-ket is designed to be as open as possi-ble, using thin but strong “arms” thatminimize early reflections at the rear ofthe cone. The area behind the spider iscompletely open as well, eliminatingair compression and “chuffing noise.”

Complete specifications for the Excelwoofer are listed in Table 1.

WHICH TL GEOMETRY?After describing the performance ofstraight TLs in the first two parts of hisseries for Speaker Builder, Augspurgerdetails five alternate geometries in Part34 that provide certain benefits over astraight pipe. Of these, two will be usedin the THOR system, and the benefits ofa third will be obtained by alternatemeans. The particular geometries arethe tapered line, the offset driver line,and the coupling chamber line. Thebenefits of each are as follows.

1. Tapering the line broadens its fun-damental resonance and thereby in-creases the frequency range of con-structive pipe output. The f3 value istypically 0.8 times fp. Attenuation ofupper harmonics is comparable to astraight line. Augspurger recommendstapers in the range of 3:1 to 4:1.

2. Offsetting the driver from theclosed end of the line by one-fifth itslength reduces the first passband dip,thus smoothing low-frequency response.However, f3 must be set about 20% high-er than fp for the flattest response.

audioXpress May 2002 9

TABLE 1EXCEL WOOFER SPECIFICATIONS

Nominal impedance 8Ω Voice coil resistance 6.1ΩFrequency range 40–2500Hz Voice coil inductance 0.4mHShort-term maximum power 250W* Bl factor 7.2N/ALong-term maximum power 100W* Free-air resonance 31HzCharacteristic sensitivity 86.5dB SPL Moving mass 15.5gVoice-coil diameter 39mm Suspension compliance 1.6mm/NVoice-coil height 16mm Suspension resistance 1.4Ns/m Air gap height 6.0mm Effective cone area 126cm2

Linear coil travel (p-p) 10.0mm VAS 37.0 ltrMagnetic gap flux density 0.88T QMS 2.5Magnet weight 0.42kg QES 0.39Total weight 1.75kg QTS 0.34*IEC 268-5

3. A coupling chamber between thedriver and the pipe inlet lowers thefundamental frequency of the combi-nation. The coupling chamber compli-ance combines with the resistiveacoustic impedance of the dampedline to produce a first-order low-passfilter that increases high-frequency attenuation.

It was clear to me at the outset that Iwanted to use a tapered line to get a

low f3 with fairly broad low-frequencyreinforcement. Additionally, with theMTM driver configuration one driver isautomatically offset from the closedend of the line. This driver offset willmitigate somewhat against the low-fre-quency extension provided by thetaper, but will help to reduce midbassresponse ripple.

Finally, folding the line provides ad-ditional high-frequency attenuation,

somewhat like that obtained with thechambered line. With these considera-tions in mind, I describe the initial lay-out and sizing of the THOR TL.

SIZING THORA big advantage of sealed and ventedbox design is that Thiele and Small,among others, have established strictrelationships between driver parame-ters and box volume and, in the case ofvented designs, box tuning, for a speci-fied frequency response. This greatlysimplifies the design process for thesesystems.

TABLE 2EXCEL TWEETER SPECIFICATIONS

Nominal impedance 6Ω Voice coil resistance 4.7ΩFrequency range 2–25kHz Voice-coil inductance 0.05mHShort-term maximum power 200W* Bl factor 3.1N/ALong-term maximum power 90W* Free air resonance 500HzCharacteristic sensitivity 88dB SPL Moving mass 0.37gVoice-coil diameter 26mm Linear voice coil travel 1.0mmVoice-coil height 1.5mm Effective piston area 7cm2

Air gap height 2.5mm Magnet weight 29gMagnetic gap flux density 0.88T Total weight 0.36kg*IEC 268-5 using 12dB/octave Butterworth high-pass filter at 2500Hz

TABLE 3Design QTS f3/fS fS/fp VAS/VpTapered 0.33 1.6 0.50 1.00(Nom. 4:1) 0.35 1.525 0.533 0.90

0.41 1.3 0.63 0.60

Offset 0.33 1.6 0.74 1.00Speaker 0.35 1.525 0.80 0.90

0.41 1.3 0.94 0.60

4"

13"6.5"

42"

15.5"

5.5"

5.5"

5.5"

11.75"

13.5"

2.75"

12.5"

8.25"9"

5"

6"

6"

25"

Centerline

86°

SIDEFRONT

38.5"

2"

6 31/32"

5.75"

4 11/32"

7/8"

3"3.5"

6 31/32"

5.75"

7/32"

¼"

7/32"

Driver

Driver

Driver

recess

recess

recess

4.5"

3"

V

V1

2

W18EX001

T25CF002

W18EX001

View through centerline

35 11/16"

4" 4"

MATERIAL:¾" MDF for walls & baffle.1" MDF Front panel and base.

FIGURE 1A: Front baffle layout for the prototype transmission lineand view through centerline.

CABINET: Bottom View FRONT

½"4½"

5¼"

5¼"

³⁄₈ "

1½"

³⁄₈"

BASE: Bottom ViewFRONT

BACK

6½"

5¼"

5¼"

1½"

13"

2½"

3"4"

1"

2³⁄₈"

Hole for driver wires

Cutout for terminals

Material for base is 1" MDF

15½"

1³⁄₈"

FIGURE 1B: Diagrams of bottomviews of cabinet and base.


B-2102-1A B-2102-1B

Subscribe to audioXpress for:Projects—including schematics, parts lists, photos, construction

details, sources—everything you need to get right to building.

Technology—theory & knowledge to improve your capabilities to

build great sounding equipment

Reviews—previews of kits—do you want to build it? Previews of

products—does it perform as the manufacturer states? Pre-

views of books and software—Is this information you want?

Letters, Tips, & Techniques—Read questions, clarifications, de-

bates, tips, and reader comments to learn even more from

each article. As a subscriber, you’re instantly part of a com-

munity of thousands of audio craftsmen like yourself.

Easy-access to information—audioXpress is written for you with

an easy-to-read table of contents and regular departments fo-

cused on specific areas of audio technology.

Information you can depend on—the knowledge presented and

the quality of articles presented in audioXpress is unparal-

leled. You’ll be treated to articles by experts in their fields in-

cluding Douglas Self, Nelson Pass, G R Koonce, and many

others from our cadre of over 700 authors.

12 monthly issues for only

$34.95!That’s less than $3.00 per

issue for all this!

PO Box 876 Dept. MDR, Peterborough, NH 03458-0876 USA

Phone: 603-924-9464 Fax: 603-924-9467 www.audioXpress.com

to request your first issue FREE TO TRY

or e-mail: [email protected]!

Ca l l 1-888-924-9465Canada add $12. Overseas rate: $59.95 for 1 year.

The significance of Augspurger’swork is that for the first time we nowhave relationships between the driverparameters fS, QTS, VAS, and the TL fre-quency, fP, and volume, VP. Strictlyspeaking, his relationships are not

unique, but they do represent an excel-lent starting point and give us con-fidence that a good design can be attained.

The starting point for TL sizing isAugspurger’s table of extended system

alignments given in Part 3 of his series.Portions of that table are reproducedhere in Table 3. These alignments areoptimum in the sense that they approxi-mate the response of an equal volumeclosed box, but with reduced cone ex-


The history of dome tweeter development at SEAS has been a longand successful one, going back more than 30 years. SEAS’ firstdome tweeter was also one of its best known and most produced.This was the original type H087, 1½″ dome tweeter, used in the leg-endary Dynaco A25 loudspeaker. This landmark loudspeaker,manufactured by SEAS, was sold in the hundreds of thousands,and served as many a budding audiophile’s introduction to truehigh-fidelity sound.

Producing the early dome tweeters was labor intensive, andconsiderable skill on the assembly line was required to produce aproduct with consistent quality. A sticky “doping compound,”used to both seal and damp the dome diaphragm, was applied byhand after the tweeter was assembled. Obviously, the amount ofthe material and the evenness of application were critical to ob-taining the desired frequency response.

Since those first designs, much research has been done to sim-plify and stabilize the process for producing soft domes. Other, non-cloth materials, such as supronyl (polyamid) plastic, were success-fully used as substitutes. But these, too, were far from ideal, becausetheir performance was highly dependent on ambient humidity andtemperature. By the late 1980s, promising new cloth materials werebecoming available which allowed the cloth to be treated prior toforming the dome. In this way, the advantages of cloth could be real-ized without the need to coat the dome after assembly.

Today, SEAS manufactures all of its dome diaphragms in house,using special vacuum-forming equipment also designed and builtby SEAS. Cloth diaphragms are produced from a proprietary mater-ial called “SONOTEX .” The SONOTEX process pre-coats the fabricfour times with a damping/sealing material, giving a nearly idealcombination of acoustic performance and high consistency.

For the Millennium tweeter, SEAS designed a special two-piecediaphragm consisting of a SONOTEX dome with a SONOMAXplastic surround. This combination results in a diaphragm withvery linear behavior and large excursion capability.

THE HEXADYM MAGNET SYSTEMThe ceramic-magnet-based magnet system found in most tweeters

has remained basically unchanged for many years. The tweeter’smagnet system performs two separate functions:

1. Supply the proper amount of magnetic flux to the voice coil2. Allow the acoustic energy generated from the rear of the dome,the surround, and the voice coil to be fully absorbed within thetweeter body without reflections, resonance, or pressure build-up.

Ceramic magnet systems are easily able to supply the necessarymagnetic energy. But they also get in the way of the rearward radi-ated energy. The construction of the system with a ring magnetcovered by top and back plates produces a large cavity in the areabetween the pole piece and the inside of the magnet.

This cavity, when excited by the movement of the voice coil, pro-duces resonance and pumping effects that will directly impact theperformance of the tweeter. Another secondary cavity sits betweenthe dome surround and the magnet system’s top plate. With suchan enclosed ceramic system, the energy build-up directly behindthe vibrating surround cannot be vented away from the magnetsystem’s top plate.

The new, patented Hexadym magnet system in the Excel Mil-lennium tweeter completely eliminates any enclosed cavities with-in the tweeter structure (Photos A and B). Instead of a single ringmagnet, the Hexadym system uses six radially magnetizedneodymium bar magnets mounted on a hexagonal pole piece. Thiscompact configuration produces large openings around the polepiece, allowing virtually all air movement generated by the di-aphragm and voice coil to be vented directly into the rear cham-ber. The Hexadym magnet system also allows airflow produced di-rectly behind the dome’s surround to be vented into the rear cham-ber through four generous openings in the top plate.

The mechanical construction of the Millennium tweeter also re-flects the no-compromise approach used in the dome and magnetassemblies. The front plate and rear chamber are constructed ofextremely rigid, die-cast zinc. This provides a virtually non-reso-nant enclosure for the tweeter, while simultaneously conductingheat away from the magnet system. John Stone

EXCEL MILLENNIUM TWEETER

PHOTO A: The Hexadym magnet system, bottom view. PHOTO B: Hexadym pole piece, top view.

cursion when the damping is adjustedfor a ±1dB passband ripple. Fortunately,the TL is a non-resonant system so thatthe optima are broad. As you will shortlysee, significant departure from the align-ments Augspurger recommends can bemade with little loss of performance.

The driver parameters needed toenter Table 3 are fS, QTS, and VAS. Val-ues of those parameters for the ExcelW18E001 woofer are given in Table 1.

The line QTS = 0.35 (which includes theeffect of the crossover coil resistanceon driver Q) in Table 3 has been inter-polated from the QTS values of 0.33 and0.41 given in Augspurger’s Table 2 ofreference [4]. Notice that the columnsheaded f3/fS and VAS/Vp are the samefor both the tapered and offset speakerconfigurations for the same QTS. Thusf3 and Vp will be the same for both, butthe line lengths will differ.

From Table 3, the predicted f3 is:

f3 = 1.525fS = 1.525 × 31Hz = 47.3Hz

For the tapered line, fp is calculated tobe:

The line length, Lp, is then one quarter

58.2Hz 0.533

31

0.533

f f S

pT ===


$129plus S&H

$99plus S&H

barrel cone truncatedcone

cube cylinder domedcylinder

truncatededge prism

4-sidedpyramid

3-sidedpyramid

truncatedpyramid

(trapezoid)

sphere wedgetruncatedcylinder

ellipsoid squareprism

opt squareprism

reg polygonprism

slantedfront prism

2-chamberBP cylinder

3-chamberBP cylinder

2-chamberBP prism

3-chamberBP prism

Design speaker boxes with Design crossovers withNeed help with your speaker designs?BassBox Pro & X•over Pro can help!TM

Sophisticated Driver Modeling and Advanced Box Design• Multiple drivers with isobaric, push-pull and bessel options. • 3 dual voice coilwiring options. • "Expert Mode" dynamically analyzes driver parameters. • Designclosed, vented, bandpass and passive radiator boxes. • "Suggest" feature providesfast box design. • All box types account for leakage losses and internal absorption.• Advanced vent calculation. • Bandpass boxes can be single or double-tuned with 2or 3 chambers. • 22 box shapes (shown below). • Open up to 10 designs at one time.• Analyze small-signal performance with: Normalized Amplitude Response, Impedance,Phase and Group Delay graphs. • Analyze large-signal performance with: Custom Ampli-tude Response, Max Acoustic Power, Max Electric Input Power, Cone Displacement andVent Air Velocity graphs. • Includes a helpful "Design Wizard" for beginners.

Design speaker boxes for any space: car, truck,van, home hi-fi, home theater, pro sound, studio,stage, PA and musical instruments. Import acousticmeasurements. For example, the response of a carcan be imported to simulate the in-car response.

Design 2-way and 3-way passive crossover net-works, high-pass, band-pass or low-pass filters,

impedance equalization, L-pads and series or par-allel notch filters. Its Thiele-Small model provides

professional results without complex testing.

Passive Crossover Filters and Scalable Driver Modeling• 1st, 2nd, 3rd and 4th-order "ladder" filter topologies. • Parallel crossover

topology. • 2-way crossovers offer Bessel, Butterworth, Chebychev, Gaussian, Legendre, Linear-Phase and Linkwitz-Riley filters. • 3-way crossovers offer All-

Pass Crossover (APC) and Constant-Power Crossover (CPC) filters. X•over Pro's capabilities scale to fit the amount of driver data. • A crossover, filter or L-pad can

be designed with just the nominal impedance of each driver. • With full Thiele-Small parameters, impedance equalization can be designed and the performance graphed.

• Graphs include: Normalized Amplitude Response, Impedance, Phase & Group Delay. • Graphs can include the full speaker response including the box. • Estimate component

resistance (ESR & DCR). • Calculate the resistance of parallel or series components.

For more information please visitour internet website at:

www.ht–audio.comTel: 616-641-5924Fax: 616-641-5738sales@ht–audio.com

Harris Technologies, Inc.P.O. Box 622

Edwardsburg, MI49112-0622

U.S.A.

Copyright © 2002 by Harris Technologies, Inc. All rights reserved worldwide. BassBox is a trademark of Harris Tech. Other trademarks belong to their respective companies. Harris Tech reserves the right to make changes without notice. All prices are in U.S. dollars.

Harris Tech Pro software for Microsoft® Windows can help you quickly create professional speaker designs. Our software is

easy to use with features like "Welcome" windows, context-sensitive "balloon" help, extensive online manuals with

tutorials and beautifully illustrated printed manuals. We include the world's largest driver database with

the parameters for many thousands of drivers.

FIGURE 2: Near-field measurements on an MTM speaker. FIGURE 3: Unfilled transmission-line impedance comparison.B-2102-2 B-2102-3

BY DENNIS COLINHere is a loudspeaker that appears to be without audible col-oration; that’s how it sounded to me. Now that’s quite a statement,deserving of intense scrutiny. After all, Joe D’Appolito designedthese speakers, so of course I thought they should be excellent.

Jiddu Krishnamurti, the (less than should be) famous observer ofthe human condition, taught that thought itself is a corruption offree observation. You must instantly forget all the past—beliefs, mem-ory, attitude, and so on, as you do when surprised by somethingnew—in order to freely and completely see the new, the present.

I have no trouble doing this when listening to (not analyzing)music I like. When reviewing an audio component, I listen; if Ihear a sonic anomaly I focus on it, then call on the analysis toolcalled thought to attempt a description, e.g., “3dB dip at 2.5kHz,”“blown woofer,” and so forth. But then I shut this off and just listenagain. After each piece of music, I write down my impressions.

You may still question whether I have a subconscious desire tobelieve a Joe D’Appolito-designed speaker must be good, or a sim-ple desire to please a master. To this I mention that I’ve also de-signed speakers, including one I want to be the world’s best. So mybiases include a competitive factor that would incline me to go theextra mile to find fault with anyone else’s speaker, including Joe’s.

But I was able to disregard all biases—positive and negative—asI sat down and just listened.

THE SOUNDCompared to any forward-only-facing speaker, I like the extra senseof ambience that a bipolar can provide, even if synthetically derived.To me, this can satisfactorily compensate the loss of original hallambience in two- (and even 5.1) channel format limitations. This,however, I find true only in a highly damped room, such as my liv-ing room. In Joe’s room, where I auditioned his speakers, there’smore liveness (although very smooth), and I think a forward-only-facing speaker is the best. My impressions are:

1. Ambience. While desiring the presence of surround speakers(which Joe provided momentarily, resulting in astonishingly good3-D ambience), this review is meant to be in pure two-channelstereo only. And as such, I heard an absolutely seamless, smooth,and deep stereo soundstage, even well off-axis. Not a hint of gaps,phasiness, or loss of tonal naturalness.

2. Tonal naturalness (“Presence” in the sonic ratings chart). Icould simply hear no flaws, not even subtle colorations! The speak-ers are so acoustically transparent, though, that I had no troublehearing many recording deficiencies, including those of the 16/44CD medium. But mind you that free of the usual speaker anom-alies, I thoroughly enjoyed the music.

One recording, even though a CD, is remarkably clean in detailand resolution: the Turtle Creek Chorale (I’ve mentioned its“goosebump factor” previously). On these speakers, I’ve neverheard more natural-sounding voice reproduction, period. The per-formers sounded right there in front of me, yet the vocal fades intoreverberation sounded infinitely far off. For a while I wanted to bea Tibetan monk so I could hear this all day!

3. Bass, midrange, treble, and balance. Sorry, nothing to com-ment on except personal perceptual flawlessness.

4. Bass. Now here’s an area for comment: These are transmis-sion-line (TL) speakers after all; TLs are supposed to have “differ-ent” bass. Before hearing these, I used to think “What’s the bigdeal? TLs are just basically highly-damped open-back cabinets,aren’t they?” Well, no!

First of all, the bass I heard was superbly damped; not a trace of“hangover” or emphasis. But the bass was also very deep and pow-erful. At one point, I was startled to hear a large bass drum impactshake me and the room down to at least 25Hz; as of this writing I

don’t know the speakers’ f3 or the bass room gain, but I do know Iheard powerful and clean output to at least 25Hz, very surprisingfrom a pair of 6½″ woofers per channel.

Second, the bass quality was even more impressive. For exam-ple, with Jacintha singing “Georgia on My Mind,” the bass viol wasthe most natural and present sounding I’ve heard (not to mentionJacintha’s voice and all else on the recording).

5. Transient Response and Image Clarity. A very good test is“Percussion Fantastic” (Fimco 017).

On these speakers, every detail of every percussion instrumentwas there with pristine immediacy and focus. With large tubularbells, for example, I clearly heard the subtle but lush midrange“knock” sound just before the blossoming resonance of the bell over-tones, all spatially and temporally correct-sounding. From the deep-est drums to what sounded like tiny (1″?) triangles (whose funda-mental was probably 5kHz), the speakers never added any confusion.

6. Overall Impression. The speakers appear to reproduce what-ever is fed them with flawless transparency. So well, they ruthless-ly reveal any recording or medium deficiency. At one point, aftercriticizing some of Joe’s recordings (which are much better thanthe average CD, mind you), I attempted to remove any sense of per-sonal offense by saying “Joe, these speakers must be first-rate toreveal such fine details of recording imperfection!”

Now, how can you argue with that if you’ve designed the speak-ers? I felt like a politician making that statement, but neverthelessit’s my true feeling.

7. One More Comment—To Sub or Not to Sub. Not! First, thesespeakers don’t need any, unless you need response down to 5Hz.Second, I don’t like separate sub (or any) woofers—I’m aware of alack of coherence on well-recorded bass transients.

With these speakers, there was no audible separation or lack ofcoherence anywhere in the audio spectrum; the response in Joe’sroom sounded flat down to 25Hz. Of course, Joe could install eight18″ woofers (per channel) into the walls and design it to extend theresponse flat to 3Hz at 130dB SPL.

OTHER SPECIFIC RECORDING IMPRESSIONSA Chorus Line—Excellent including bass; opening percussion withgreat depth estimated to 25Hz (with possible room-gain contribu-tion).

Carmen—Reproduced very clearly at peaks above 100dB SPL;speakers had no problem with the 200W/per channel amplifierprobably driven near clipping.

Beethoven “Pastoral”—Tonality good, but recording seemed tohave constricted ambience.

Fanfare for the Common Man—Very natural, deep, and spacious.Chopin/Rubenstein—Hauntingly good music; piano recording

technique sounded somewhat distant and dull. Not from speakers;other piano recordings could be first-rate.

Tannoy Hi-Fi Sound Sampler—One of the best recordings ofstring bass. Speakers delivered perfect-sounding tonality andspace/time coherence. The overtone structure, instrument reso-nance, and sense of live string pulsiness sounded not just sepa-rately—good, but coherent live instruments.

Dvorak Symph. #9, Solti—Absolutely excellent in all regards.BSO, von Stade—Very good instruments, voice somewhat over-

loaded in recording (present at any playback level, plus if it werethe speakers, the instrument sound would have been intermodu-lated with the voice; it wasn’t).

Dvorak Piano Quartet—No problem with this piano recordingand reproduction thereof! One of the most analog-sounding (ormore correctly, non-digital-sounding) CDs I’ve heard.

Up to this point, I hadn’t seen any measurements. But now it’stime to open the secret envelope. (to page 16)

LISTENING CRITIQUE


of a wavelength at fpT.

In this equation, c, the speed ofsound in air, is taken to be 13584 in/sec.A similar calculation for the offset linegives a line frequency of 39.2Hz and aline length of 86.6″ . The total line

length actually used is 81″. With thislength an f3 of 44Hz is achieved.

Column 4 of Table 3 tells how to com-pute the TL internal volume as a func-tion of VAS. From Table 3 you find that:

Now VAS for the MTM configuration is

twice the VAS of a single driver or 74L.Solving for Vp you get:

You must establish at least two di-mensions of the enclosure before youcan use the volume calculation to getthe final dimension. Here some practi-

33ASp in 5021 2.905ft 82.2L

0.9

74

0.9

V V =====

0.9 V

V

p

AS =

in 58.4 58.2 4

13584

f

c L

ppT =

×==


FIGURE 4: TLine woofer impedance with 13 oz filling. FIGURE 5: TLine responses with 13 oz filling.

CT101 key specificationsGain (selectable) 0, 6 or 12 dB

25 MHzSlew rate (at 0dB gain) 500 V/uSS/N ratio (IHF A) 112 dBTHD 0.0002 %Output resistance 0.1 ohmChannel matching ± 0.05 dBPCB dimensions: 100 x 34 mm

3.97 x 1.35 "

General attenuator specificationsNumber of steps: 24Bandwidth (10kOhm): 50 MHzTHD: 0.0001 %Attenuation accuracy: ±0.05 dBChannel matching: ±0.05 dBMechanical life, min. 25,000 cycles

Fax: (+66) 2 260 6071 E-mail: [email protected]

gwith a stereo CT1 attenuator added.

CT100 key specificationsGain (selectable): 40 to 80 dBRIAA eq. deviation: ± 0.05 dB S/N ratio (40/80dB gain): 98/71 dB THD: 0.0003 %Output resistance: 0.1 ohmChannel separation: 120 dBBandwidth: 2 MHzPCB dimensions: 105 x 63 mm

4.17 x 2.5 "

CT2 6-gangvolume control for A/V Audio

B-2102-4 B-2102-5

cal considerations entered. I knew I wanted a tweeter height of

34–35″ to match ear height at my fa-vorite listening position. I also knewfrom earlier experience with woofers ofthis size that a front baffle width should

be no greater than 9″ for uniform hori-zontal polar response. Finally, I wantedto isolate the crossover from the mainacoustic volume by placing it in thebase of the enclosure.

These considerations led to a trial

layout for the front baffle of the THORTL shown in Fig. 1A. Now all that wasneeded was to determine the interiorand exterior depths of the line. A firstcut at line depth went like this. Assum-ing 0.75″ MDF for the sides and topleads to an internal width and height of7.5″ and 41.25″, respectively.

The internal depth, d, is then com-puted as follows:

To get the external depth you must addthe thickness of a 1″ front baffle, a 0.75″internal baffle, and a 0.75″ rear panelfor an overall depth of 18.75″. This num-ber was considerably deeper than Iwanted and would lead to a rather largeand heavy enclosure.

in 16.25 41.25 7.5

5021 d =

×=


(from page 14)

COMMENTS ON MEASUREMENTS1. Joe has often said “Horizontal frequency response over a 60° arcis a good measure of perceived frequency response.” Suffice it tosay that Fig. 16 agrees with my perception.

2. Regarding bass extension—Figure 13 shows an LF −3 of 44Hz,with an ultimate LF slope of about 12dB/octave (similar to aclosed, not vented box). As previously mentioned, I heard what Iestimated to be strong 25Hz output. Figure 13 is down about 12dBat 25Hz, so I would say room gain is helping here.

Since room gain (called “cabin gain” in a car) boosts LF output(re free air) at 12dB/octave, this can very well compensate a speak-er’s 12dB/octave rolloff. This is also true of closed-box systems, butthe THOR TL had absolutely no “box-like” sound; bass was su-perbly natural.

ABOUT ROOM GAINIn free space, a small (re bass wavelengths) source must deliverconstant air acceleration to radiate a constant sound pressurelevel versus frequency, because at lower and lower frequencies,less and less of a wavelength is “grabbed” by the source. Below f3of a closed-box speaker, however, the cone excursion (and air-vol-ume displacement) is constant; acceleration (thus SPL) falls off at12dB/octave. Figure 13 shows THOR to do this also.

But in a confined space (room), below the frequency where thelongest room dimension is about half wavelength, the air becomespressurized as a whole. Thus, a constant speaker volume displace-ment produces air pressure cycles (SPL) of constant amplitude ver-sus frequency; with no leakage (room and speaker), this would ex-tend all the way to DC. So eight 18″ woofers mounted in a wallcould produce 130dB SPL at 3Hz—not recommended if you valueyour hearing (and walls)!

RECORD REFERENCES1. Chesnokov: “Spaseniye sodelal,” Turtle Creek Chorale, TimothySeelig, cond., Track 3, HDCD Sampler, Reference Recording RR-905CD.2. Hamlisch/Morita: “A Chorus Line,” Turtle Creek Chorale/DallasWomen’s Chorus, Timothy Seelig, cond., Track 4, HDCD Sampler,Reference Recording RR-905CD.

3. Jacintha, “Georgia on My Mind” from “Here’s to Ben,” First Im-pressions Music, FIM XRCD 020.4. “Carmen Fantasy” from “Percussion Fantasia,” Harold Faber-man and the All Star Percussion Ensemble, First Impressions Music, FIMCD 017 (HDCD 24-bit recording).5. Bizet-Shchedrin, “The Carmen Ballet,” Orchestre Philhar-monique de Monte Carlo, James DePriest, cond., Delos 3208.6. Beethoven, Symphony No. 6, “The Pastoral,” Hanover Band, RoyGoodman, cond., Nimus Records, N15099.7. Chopin, “The Nocturnes,” No. 2 in D-flat major, Artur Ruben-stein, Piano, Musical Heritage Society 523870T.8. Tannoy HiFi Series Sound Sampler, band 11, “Im Uomini, InSoldati,” Mozart, Cecilia Bartoli, Wiener Kammerorchester, GyOr-gy Fischer, cond.9. Dvorak, Symphony No. 9, “From the New World,” Chicago Sym-phony Orchestra, Sir George Solti, cond., London 410 116-2.10. Berlioz, “Nuits d’ete,” Frederica Von Stade, BSO, Seiji Ozawa,cond., CBS Master Works, MK 39098.11. Dvorak, Piano Quartet in E-flat Major, OP. 87, 2nd movement,The Ames Quartet, Dorian-90125.

SONIC CHARACTERISTICS RATINGS

DCPresence

DCTreble Quality

DCMidrange Quality

DCBass Quality

DCBass Extension

DCImmed. & Trans Response

DCImage Focus

DCStereo Soundstage Realism

DCAmbience

DCFreedom From Distortion

DCF.R. Smoothness

DCL-M-H Balance

1 2 3 4 5 6 7 8 9 10

PHOTO 2: Prototype crossover.

At this point I made a number of arbi-trary decisions. I chose a line taper of3:1 and limited the overall depth to13.5″. This led to the internal layout ofthe line shown in Fig. 1A. Placing the in-terior baffle at an angle produces the de-sired taper. A side, but important, bene-fit of the interior baffle is that it addsgreatly to enclosure rigidity, effectivelyclamping the side panels together andlargely eliminating side-wall vibration.

The resulting layout has a throatarea of 61.875 in2, which is roughly 1.6times the combined diaphragm area ofthe two drivers, and an exit area of20.625 in2. As you will see, this depar-ture from Augspurger’s recommenda-tion has little effect on f3.

FILLING THE LINEIn Table 1, Part 3, of the Augspurger ar-ticle4, the author recommends packing

densities for four filling materials. Theoptimum packing density is a functionof line length. For my first trials I usedpolyester fiber (“Poly-Cat” polyfill avail-able at Wal-Mart).

For a line length of 81″ the recom-mended packing density is 0.78lbs/ft3. To get the total amount of poly-ester needed you must calculate theTL internal volume. From Fig. 1A theinternal volume is calculated to be


FIGURE 6: TLine woofer impedance with 26 oz filling. FIGURE 7: TLine responses with 27 oz filling.

ACCUTON

A Proud North American Distributor of

P.O. Box 9085, Wichita Falls, TX 76308, USA

Phone: (940) 689-9800; Fax: (940) 689-9618; E-mail: [email protected]

See us at: www.soniccraft.com

B-2102-6 B-2102-7

1.934ft3, and the total amount of poly-fill needed is then 1.934 × 0.78 = 1.61lbs = 24.1 oz.

In the past, some authors have sug-gested varying the packing densityalong the line length, but Augspurgerfound no particular advantage to this inhis studies and recommended a uni-form density. Getting a uniform pack-ing density is a bit tricky, however, be-cause the line volume per unit length ischanging due to the taper.

Referring again to Fig. 1A, the vol-umes of the two line sections, V1 andV2, are found to be 1.208ft3 and 0.725ft3,respectively. These volumes represent62.5% and 37.5% of the total line vol-ume, respectively. Now to get a uni-form packing density, you shouldplace approximately 15 oz of the poly-fill in V1 and 9 oz in V2.

THE APPROACH TO LINE TRIMMINGOnce TL dimensions are set, final trim-

ming of the TL packing density is doneusing a sequence of electrical imped-ance and acoustic measurements. Icould jump to the final result, but Ithink the various steps in the processare instructive because they can beused in general to trim any transmis-sion line. I will also take this opportuni-ty to compare results from the taperedline with an equivalent straight line dri-ven with a single woofer.

The acoustic measurements are simi-lar to those used by Augspurger in hisarticle. Near-field woofer and port SPLmeasurements are taken using theCLIO measurement system in the MLS

mode. The near-field technique is usedto overcome the effects of low-frequen-cy standing waves.

In this technique, the microphone isplaced very close to the driver di-aphragm (<0.1″) to swamp out diffrac-tion and room effects. At low frequen-cies where the diaphragm acts like arigid piston, the measured near-fieldresponse is directly proportional to thefar-field response and independent ofthe environment into which the driverradiates. Based on the diameter of theW18 woofers, the near-field woofermeasurements are valid up to about860Hz.

TABLE 4THOR SYSTEM SPECIFICATIONS

Frequency range(Hz) 40–25000 Sensitivity 89dB SPLShort-term maximum power* 400W Recommended amplifier 50–400WLong-term maximum power* 200W Dimensions(mm) 229 × 1060 × 343Crossover frequency 2500Hz Bass loading Transmission lineImpedance 4Ω*IEC 268-5

FIGURE 10: Woofer pair and tweeter impedance.FIGURE 8: Woofer-port transfer function comparison.


FIGURE 9: Woofer pair and tweeter frequency responsesabove 100Hz.

PHOTO 3: Prototype TL.

B-2102-8

B-2102-9

B-2102-10

CLIO works in the time domain andproduces both amplitude and phase re-sponse data. The woofer and port re-sponses are measured separately andthen added, taking proper account ofphase and woofer/port area differencesto get the complete low-frequency re-sponse of the line. This process is de-scribed in detail in Chapters 4 and 7 ofmy book, Testing Loudspeakers5. Photo1 shows the lab setup for testing andtrimming the line.

When measuring the port near-fieldresponse, you must place the micro-phone in the plane of the port exit.The port measurement is then correct-ed by multiplying it by the square rootof the ratio of port area to the com-bined area of the two woofers. Thiscorrection is:

After correction, the port response isadded to the two woofer responses to get the complete near-field TL response.

0.727 0620.625/39.

correction response port

=

=


.

.

. .

.

..

.

.

.

.

FIGURE 11A: Crossover network for THOR and ODIN Mk3.

B-2102-11A

There is a potential problem withnear-field measurements of woofer re-sponse with the MTM configuration. Ifboth woofers are driven simultaneous-ly, the near-field response of one woofer

can be contaminated by the outputfrom the second woofer because theyare so close together. This is illustratedin Fig. 2.

The results shown in Fig. 2 are for an

MTM speaker using two 5.25″ woofersin a sealed enclosure, but the resultswill apply equally well to THOR. In thisseries of tests, the microphone isplaced about 0.1″ in front of the upperwoofer dust cap.

Figure 2 has three plots. One plot isthe near-field data taken at the upperwoofer with the lower woofer terminalsshorted. A second plot shows theacoustic response measured at theupper woofer with the lower woofer driven and the upper woofer shorted.Below 60Hz, the lower woofer signal is only 10dB below the upper wooferoutput.

The third curve shows the responseat the upper woofer with both woofersdriven. You see that this plot is contam-inated with some of the output from thelower woofer. In practice, the near-fieldresponse of each woofer is measuredseparately with the other woofer short-ed, and then both responses are addedto get the total low-frequency responseof the woofer pair.

So far I’ve spoken only about acoustic

FIGURE 12: Crossover voltage responses. FIGURE 14: System and individual driver responses.

FIGURE 13: Full-range on-axis frequency response.

L1

L2 C1

C2

C3

R2

R1 R3

L3

C4

+ +

+

-

-

-

+

IN IN

OUT

-

OUT

1mH

0.5 .15mH 8.2µF

0.24mH

8µF

18µF

2.2µF

12.5

2R

FIGURE 11B: Crossover layouts.

FIGURE 15: Horizontal polar response.


B-2102-11B

B-2102-12

B-2102-13 B-2102-15

B-2102-14

measurements, but you can also tell agreat deal about TL performance fromimpedance data. Figure 3 compares themeasured woofer impedance of the ta-pered, folded line with an equivalentstraight line. Both lines are unfilled.The straight-line plot is offset by +4Ω toease the comparison.

First notice that both lines exhibitthe double-peaked curve of a ventedloudspeaker. Thus, up to first order, theunfilled line acts much like a bass re-flex speaker. However, you also see ad-ditional impedance peaks due to high-er-order modes of the line. Beyond thefirst two peaks, the straight-line TLshows four additional peaks with in-creasing frequency.

Contrast this with the tapered, fold-ed line where: 1) the minimum be-tween the two lower peaks occurs at alower frequency, 2) the curve about theminimum is broader and shallower, 3)all peaks are more highly damped rela-tive to the straight line, and 4) only atotal of three peaks are seen in thefolded, tapered line versus six for thestraight line.

These results support the contentionthat a tapered line has a lower funda-mental resonant frequency and abroader range of support. The absenceof higher-frequency peaks is due tofolding the line. You will see this moreclearly later on when you comparetransfer functions for optimallydamped straight and tapered lines inFig. 8.

TRIMMING THE LINERather than going directly to the “opti-mum” calculated packing density, Ifirst packed the line uniformly to halfthe recommended density; i.e., 13 oz ofpolyfill. The impedance plot for thiscondition is shown in Fig. 4. Notice thatthe first impedance peak just below20Hz is almost gone. The same is truefor the third peak just above 100Hz. Theimpedance curve looks almost like thatof a closed box speaker.

Responses of the woofer pair, theport, and their sum are shown in Fig. 5.The summed response shows a peak-to-peak ripple below 500Hz of ±1.7dB. Thelow-frequency f3 point relative to 500Hzis 41Hz. Below 41Hz response falls offat 12dB/octave.

Finally, observe that the port outputaugments woofer output by 4–5dB at allfrequencies between 20Hz and 110Hz.From these results you can concludethat the lightly damped TL acts like anunderdamped closed box system with4–5dB increased low-frequency outputcapability.

The impedance curve of the optimal-ly filled TL (24 oz) is shown in Fig. 6.Now all traces of line modes are goneand the curve is almost purely second-order like that of a closed box. The lineis now essentially non-resonant. Re-sponses of the woofer pair, the port,and their sum are shown in Fig. 7. Thesummed response ripple is ±0.6dB andf3 is 44Hz. Bass augmentation averages3–4dB from 20Hz to 100Hz.

From these results you can concludethat for a fixed line length there is atrade-off between ripple response and f3controlled by line damping. You alsosee that the line can be damped effec-tively by observing only the impedancecurve. Damping should be adjusteduntil all traces of line modes just disap-pear from the impedance curve.

From the impedance plot you cancompute an equivalent QTC for a sec-ond-order system with the same imped-ance curve using any of the proceduresoutlined in Chapter 2 of reference [5].The value obtained is QTC = 0.55, indi-cating that the woofer pair is almostcritically damped.

The woofer/port transfer functionplot shows the acoustic output at theTL port produced by the acoustic inputto the line coming off the rear of thewoofer cones. If you compare thewoofer/port transfer function for an optimally damped straight line againstan optimally damped tapered and fold-ed line, you get the plot shown in Fig.8. Between 100 and 400Hz and againabove 700Hz there is much less high-frequency acoustic output from theport of the folded, tapered line. Thisgreatly reduces ripple in the 100 to400Hz range relative to that of astraight line.

DESIGNING THE CROSSOVERWith the line optimally damped, my ef-forts now turned to the design of thecrossover. Crossover design for me is athree-step process. First, I placed all


Madisound Speaker Components, Inc.P.O. Box 44283

Madison, WI 53744-4283 USATel:608-831-3433 Fax:608-831-3771

[email protected]

seas THORT-Line Speaker Kit

THOR is a new top-of-the-lineloudspeaker kit developed by Dr.Joseph D’Appolito for SEAS. It isa full range, floor-standing systemthat provides the very highest levelof acoustic performance.

This kit uses two Seas ExcelW18E001 magnesium woofers andT25CF002 Millennium Sonotexdome tweeter, mounted in aD’Appolito configuration. Theenclosure uses a newly developedTransmission Line configuration,derived from sophisticatedcomputer modeling and extensiveexperimentation. Usable in-roombass response extends to 30Hz.

Thor is an audiophile product sureto impress discriminating listeners.The highs are transparent. Themidrange has exceptional clarityand openness. The bass is deepand solid.

The cabinets are assembled Oakveneer with solid Oak corners, in aClear or Black finish. The grills areblack and are recessed flush withthe speaker front. Assemblyrequires some skill at soldering.

Please visit our web site for moredetailed information and upgrades.

Price per pair: $1585.00

Dimensions: 45½” H x 9” W x 14¼” D

drivers in the prototype enclosure ofFig. 1A and made acoustic and electri-cal measurements on them. The mea-surements include acoustic frequencyand phase response, acoustic phasecenter and electrical impedance. Thisprocess is described in detail in Chap-ter 7 of reference [5].

I then enter this data into one of themany crossover optimization pro-grams I have to develop a preliminarycrossover design. Lest you think thatthis process is automatic and that thesoftware does all the work, be warnedthat these optimization programs arequite dumb. They cannot decide on anoptimum crossover topology and theydo not know which components shouldbe optimized and which should be leftalone. This is where the “art” ofcrossover design with optimization soft-ware comes in. The software savesmany hours of experimentation, pro-ducing a preliminary design that getsyou quickly into the ballpark, but thedesigner must pick the right crossovertopology and guide the optimization

process to a reasonable result.In the last step I built the prelimi-

nary crossover and auditioned it exten-sively, and used these listening tests forthe final tailoring of loudspeaker per-formance.

CROSSOVER DESIGN CRITERIAIn designing a crossover I have two pri-mary requirements: 1) flat on-axis firstarrival response and 2) uniform hori-zontal polar response. Directional cuesso important to imaging are deter-mined primarily by a loudspeaker’sfirst arrival response, which should berelatively flat to avoid amplitude distor-tion of the directional information.

However, the overall frequency bal-ance of a loudspeaker as perceived bya human listener is a combination ofdirect and reflected sound. Off-axis en-ergy arrives at the listening positionafter reflection off the walls. In typicallistening rooms this energy arriveswell within the Haas fusion zone, atime interval starting just after thefirst arrival and extending out to

40–50ms. Even if the on-axis responseis flat, poor off-axis response can pro-duce a perceived colored frequencybalance.

For good stereo imaging and properspectral balance from side-wall reflec-tions, the horizontal polar response off-axis curves should be smooth replicasof the on-axis response with an allow-able exception for the natural rolloff ofthe tweeter at higher frequencies andlarger off-axis angles. (Our ear-braincombination tends to reject higher-fre-quency side-wall reflections.)

There are several other importantquantitative measures of speaker per-formance, but these are not controlleddirectly by the crossover network. Seemy many loudspeaker test reviews inaudioXpress for a complete discussionof these other measures.

INDIVIDUAL DRIVER TESTINGFigure 9 shows the quasi-anechoic fre-quency response (first arrival response)of the woofer pair and tweeter with themicrophone placed on the tweeter axis

FIGURE 18: TLine responses: 13 oz front and 7 oz batting.

PHOTO 4: Parts kit from Madisound.

FIGURE 16: THOR average horizontal response over ±30°.

FIGURE 17: THOR impedance magnitude and phase.


B-2102-16

B-2102-17

B-2102-18

at a distance of 1.25m. The plot scalecovers 100Hz to 20kHz. The data is thennormalized to 1m to get driver sensitivi-ty. (Woofer pair response below 100Hzis determined via near-field techniquespreviously discussed.)

Tweeter response averages 90dBSPL/1m/2.83V above 2kHz. Below 2kHztweeter response falls off smoothly witha slightly over-damped response.

Starting at about 1500Hz the wooferpair response falls 5dB with decreasingfrequency, reaching a uniform level ofabout 90dB at and below 400Hz. Thefall-off is due to the spreading loss char-acteristic of all woofers on narrow baf-fles ([5], Chapter 4). The woofer peaks to100dB at 4.4kHz and then falls off at anaverage rate of 24dB/octave one octaveabove that frequency.

Frequency responses of the wooferpair and tweeter overlap between 1.2kHzand 5kHz, suggesting that a preliminaryvalue of 2.5kHz for the crossover fre-quency would be a good place to startdesign. This frequency may be subject tochange depending upon the resultinghorizontal polar response.

Woofer pair and tweeter impedancesare plotted in Fig. 10. The woofer pairimpedance of Fig. 7 has been extendedout to 20kHz.

CROSSOVER TOPOLOGY SELECTIONAND OPTIMIZATIONI favor in-phase, i.e., even-order,crossovers for most applications be-

cause they are the least sensitive tointer-driver phase differences and tim-ing errors. In the case of the MTM con-figuration they also limit off-axis re-sponse in the vertical which greatly re-duces floor and ceiling reflections. Forthe THOR TL the goal was to design afourth-order acoustic crossover re-

sponse at a crossover frequency of2500Hz.

The woofer crossover must accom-plish three functions: 1) control the re-sponse rise between 400Hz and 1.5kHz,2) suppress the 100dB woofer peak at4.4kHz, and 3) provide the final high-frequency rolloff, which, when com-


Precision Acoustic MeasurementsRequire Precision Microphones

PS9200KIT™ $1650 USDA complete IEC and ANSI traceable Type 1 MeasurementMicrophone System 2 Hz to 40 kHz, 15 dBA to 160 dBSPL

*½ inch capsule *4012 Preamp *PS9200 2 Channel PS*AC adaptor *WS1 windscreen *SC1 die cut storage case.Options: 511E Calibrator; 1 & ¼ inch mics; and Gain for

DRA’s MLSSA and other board level products.

ACO Pacific, Inc.2604 Read Ave. Belmont, CA 94002 USATel: (650) 595-8588 FAX: (650) 591-2891

e-mail [email protected]

ACOustics Begins With ACO™

bined with the woofer’s natural re-sponse, produces the desired 24dB/oc-tave acoustic decay.

To adequately protect the tweeter theelectrical rolloff of the tweetercrossover must be at least 12dB/octaveat all frequencies below crossover. Thetweeter’s acoustic rolloff below 1kHzbegins at a rate of 6dB/octave and tran-sitions to 12dB/octave below 300Hz.The tweeter crossover must thereforeproduce an electrical attenuation start-ing at 18dB/octave and then transition12dB/octave to attain the desired over-all 24dB/octave acoustic rolloff.

With these points in mind thecrossover circuit topologies I finally

selected are shown in Fig. 11. Look atthe woofer pair crossover first. Thereis a tendency in crossover design toseparate the basic crossover actionfrom the specialized functions ofspreading loss correction and re-sponse peak suppression.

There is also an often-unthinking useof Zobel impedance compensationwhen better performance is often obtained without one. This leads tooverly complex crossovers. The woofercrossover topology I finally settled oncombines the three required functionswith an economy of parts and results inabsolutely astounding and seamlessdriver integration.

Woofer and tweeter crossover volt-age transfer functions after optimiza-tion are shown in Fig. 12. For those ofyou with some circuit theory back-ground, the woofer crossover is a third-order electrical filter with a second-order zero. The woofer crossover volt-age response is explained as follows.

L1 provides an initial rolloff of6dB/octave starting at 400Hz to compen-sate for rising response of the wooferpair. The R1, C1, L2 triple forms a series-resonant shunt that comes into playaround 2500Hz. It produces a 31dBnotch at the woofer peak and providesadditional high-frequency rolloff. Resis-tor R1 controls the depth of the notch.Finally, beyond 10kHz the woofercrossover response flattens out, but thatis OK because the woofers are falling offat 24dB/octave above the notch.

With a Zobel, the woofer crossoverresponse would continue to fall offabove 10kHz at a 6dB/octave rate. With-out a Zobel, however, the rising imped-ance of L1 is matched by the rising im-pedance of the woofer pair voice coils,resulting in no net electrical rolloff.

Figure 12 also shows the 18dB/octaverolloff required by the tweeter below1kHz. The tweeter crossover output isdown 36dB at 600Hz, the tweeter’s mea-sured resonant frequency. The transi-tion to 12dB/octave occurs below thescale of the plot.

Crossover parts values are also listedin Fig. 11A. It is very important to usethe specified coil wire size for L3. Below300Hz L3 coil resistance dominatesover coil inductive reactance so thatthe crossover looks like a double RC fil-

ter giving the required 12dB/octave at-tenuation. A larger wire size would re-duce coil resistance and push the tran-sition frequency down to a lower value.Resist the urge to use a larger wire size.Photo 2 shows the prototype crossover.

FREQUENCY AND POLAR RESPONSE TEST RESULTSPhoto 3 shows the prototype TL readyfor testing in my lab. Figure 13 showsthe full-range quasi-anechoic frequencyresponse obtained with the micro-phone placed on the tweeter axis at adistance of 1.25m. Response is flat with-in ±1dB from 200Hz to 20kHz. Low-fre-quency f3 is 44Hz. Sensitivity averages88dB SPL/1m/2.83V.

Figure 14 shows system frequency re-sponse and response of the individualdrivers on an expanded frequency scale.On this plot the crossover frequency ishighlighted at 2526Hz, satisfyingly closeto the target crossover of 2500Hz.

Horizontal polar response is exam-ined in Figs. 15 and 16. Figure 15 is awaterfall plot of horizontal polar re-sponse in 10° increments from 60° right(−60°) to 60° left (+60°) when facing thespeaker. All off-axis plots are refer-enced to the on-axis response, whichappears as a straight line at 0.00°. Thus,the plotted curves show the change inresponse as you move off-axis.

For good stereo imaging the off-axiscurves should be smooth replicas of theon-axis response with the possible ex-ception of some tweeter rolloff at high-er frequencies and larger off-axis an-gles. For home theater applications amore restricted high-frequency re-sponse may be desirable.

From Fig. 15 you find that the −3dBbeam width at crossover is ±50°. There isa bit more off-axis droop around 1500Hz,but the −3dB beam width is still ±45°.Above 15kHz and at angles greater than40° there is a fairly steep fall-off in re-sponse that is characteristic of 28mmtweeters with a recessed dome. But, as Iindicated earlier, this performance isperfectly acceptable. The −3dB beamwidth at 15kHz is still ±25°.

The average horizontal frequency re-sponse over a 60° arc is a good measureof perceived frequency response. Thisaverage response is plotted in Fig. 16.Relative to 1kHz, response at 10kHz is


PHOTO 5: An enclosure version fromMadisound.

FIGURE 19: Enclosure plans.

B-2102-19

down only 0.9dB. At 20kHz the figure is1.4dB. This plot, in particular, showsTHOR’s excellent in-room frequencybalance.

THOR’s impedance magnitude andphase are plotted in Fig. 17. The mini-mum impedance of 3.6Ω occurs at180Hz. The impedance peak of 18.3Ωat 1.5kHz is caused by the interactionof the woofer and tweeter crossovernetworks forming a parallel resonanceat that frequency. The maximumphase angle of 45° occurs at 2140Hz,but the impedance magnitude at that

point is 10Ω. The system impedance israted at 4Ω.

PRACTICAL CONSIDERATIONSAfter many months of operation, theDacron pillow filler settled in the sec-ond half (the rising part) of one of thelines. This occurs only in the secondhalf of the line because it expands to-ward the bottom of the enclosure givinglittle support to the filling material. Thesettling did not appear to affect perfor-mance, but the problem can be avoidedaltogether by using either Acousta Stuf

(available from MahoganySound) or Dacron Quiltpadding in the second half ofthe line. Performance will bethe same with either solution.

In the case of Acousta Stufyou will need 21 oz of materialdivided into 13 oz for the firsthalf and 8 oz for the secondhalf of the line. This materialmust be thoroughly teased outto fill each volume.

Alternatively, you can fill thesecond half of the line with

Dacron Quilt padding, which will retainits shape when placed in the line. Youwill need about 9 oz of the material. Cutit into three 7.5″ wide strips.

The first strip should equal thelength of the last half of the line. Thesecond and third strips should be two-thirds and one-third the length of thefirst, respectively. The longest strip fillsthe second half of the line, while thesecond and third strips fill two-thirdsand one-third of the lower portions ofthe line, respectively. Low-frequencyresponse using the quilt padding isshown in Fig. 18.

CONSTRUCTIONI will not give detailed instructions forbuilding the THOR enclosure. Enclo-sure plans are given at the end of thisarticle (Fig. 19) and also are availableon the SEAS website at www.seas.no.We have provided a cutting guide (Fig.20) that also specifies the total amountof material needed for each enclosure.Any experienced woodworker shouldbe able to follow the plans without dif-ficulty.


4470 Avenue ThibaultSt-Hubert, QC J3Y 7T9Canada

SOLEN INC. Tel: 450.656.2759

Fax: 450.443.4949

Email: [email protected]

WEB: http://www.solen.ca

[ y

COMPONENTS:

SOLEN HEPTA-LITZ AND STANDARDINDUCTORS ANDCAPACITORS - THECHOICE OF MANYHIGH-END SPEAKERMANUFACTURERS.

HARDWARE:

POWER RESISTORS, L-PADS, CABLE,ABSORBING ANDDAMPING MATERIALS,GOLD SPEAKERTERMINALS, GOLDBANANA PLUGS ANDBINDING POSTS, GRILLFASTENERS, PORTTUBES AND TRIMRINGS, PAN HEADSCREWS, SPIKES AND TEE NUTS WITHALLEN HEAD BOLTSAND PLENTY MORE...

Solen crossover components - used by the most discriminating

loudspeaker manufacturers.

SOLEN HEPTA-LITZ INDUCTORSAir Cored Inductors, Litz-Wire Perfect Lay Hexagonal WindingValues from .10 mH to 30 mHWire Size from 1.3 mm (16AWG) to 2.6 mm (10 AWG) 7 Strands

SOLEN STANDARD INDUCTORSAir Cored Inductors, Solid Wire Perfect Lay Hexagonal Winding

Values from .10 mH to 30mHWire Size from 0.8 mm (20AWG) to 2.6 mm (10 AWG)

SOLEN FAST CAPACITORSFast Capacitors, Metalized PolypropyleneValues from 0.10 µF to 330 µFVoltage Rating: 630, 400, 250 VDC

CROSSOVER AND SPEAKER PARTSMetalized Polyester Capacitors, 1.0 µF to 47 µF, 160 VDC, Non PolarElectrolytic Capacitor, 22 µF to 330 µF, 100 VDC, Power Resistors 10 W, 1.0 Ω to 82 Ω, 8 Ω L-Pads plus all the hardware and suppliesto complete any speaker project.

C A L L T O O R D E R T H E S O L E N C R O S S OV E RC O M P O N E N T C ATA L O G F O R $ 8.00 PACKAGEDWITH T H E C O M P R E H E N S I V E S O L E N S P E A K E RD R I V E R C O M P O N E N T S C ATA L O G .

PHOTO 6: Madisound’s finished crossover.

For those of you who do not care tobuild the cabinets from scratch, enclo-sures are available from the sourceslisted at the end of the article. A com-plete kit of parts including drivers andcrossovers is also available from thesesources. Photo 4 shows the parts kitprovided by Madisound. One version ofthe enclosure also available fromMadisound is shown in Photo 5, filledand ready for driver installation. Photo6 shows a finished crossover mountedin the base of the enclosure.

SUMMARYIn this article you have seen that

Augspurger’s work represents an ex-cellent starting point for the design oftransmission-line loudspeakers. Hisrecommendations on packing densityversus line length are right on target.Once a prototype line is built, the opti-mum packing density is easily deter-mined experimentally with a se-quence of acoustic and/or electricalimpedance measurements. Similaracoustic and impedance measure-ments on the drivers mounted in theprototype enclosure then provide thedata for rapid CAD design of a trialcrossover network.

REFERENCES1. G. L. Augspurger, “Loudspeakers in DampedPipes—Part One: Modeling and Testing; and PartTwo: Behavior,” 107th JAES Convention, 24–27 Sep-tember, 1999, Preprint No. 5011.

2. G. L. Augspurger, “Transmission Lines Updated,Part 1,” SB 2/00.



5. J. A. D’Appolito, Testing Loudspeakers, Audio Ama-teur Corporation, Peterborough, NH, 1998.

SOURCES FOR THOR KIT PARTSZalytron Industries: www.zalytron.com

Madisound: www.madisound.com

12.5" 12.5" 12.5" 12.5" 7.5" 7.5" 7.5" 7.5" 9"

Baffle Top

Top

BaffleBackBackSide 1 Side 1Side 2 Side 2Rt Rt Rt Rt Rt

Rt Rt

Lft Lft Lft Lft

Lft

LftLft

41¼" 38½

35 1

1/16

"

12½"

4¼"

4¼"

Corners (4)Cut 7.5" edges @ 45°

4x8' x ¾" MDF CUTTING GUIDE

9" 9" 13"

13.5"

3"

3" 3"

3"

3"

3"

13.5"

15½"

15½"42"

Front FrontRt Lft

Base Top

Base Top

Rt

Lft

Base Sides

Base Fronts& Backs

4 x 8' x 1" MDF CUTTING GUIDE

42"

NOT TO SCALE

FIGURE 20: Cutting guide.


B-2102-20


LANGREX SUPPLIES LTDDISTRIBUTORS OF ELECTRONIC VALVES, TUBES & SEMICONDUCTORS AND I.C.S.

1 MAYO ROAD, CROYDON, SURREY, ENGLAND CR0 2QP24 HOUR EXPRESS MAIL ORDER SERVICE ON STOCK ITEMS

E-MAIL: [email protected]

A SELECTION OF OUR STOCKS OF NEW ORIGINAL VALVES/TUBES MANY OTHER BRANDS AVAILABLE

MANY OTHER BRANDS AVAILABLEThese are a selection from our stock of over 6,000 types. Please call or FAX for

an immediate quotation on any types not listed. We are one of the largest distributors of valves in the UK. Same day dispatch. Visa/Mastercard acceptable.

Air Post/ Packing (Please Enquire). Obsolete types are our specialty.

5R4GY RCA 7.505U4GB SYLVANIA 10.005Y3WGT SYLVANIA 5.006BX7GT GE 7.506FQ7 SYLVANIA 7.506L6GC SYLVANIA 20.006L6WGB SYLVANIA 15.006SL7GT USA 5.006SN7GT USA 7.506V6GT BRIMAR 7.5012AX7WA SYLVANIA 6.0012BH7 BRIMAR 12.0012BY7A G.E. 7.00211/VT4C G.E. 85.00807 HYTRON 7.505687WB ECG 6.006080 RCA 10.006146B G.E. 15.006973 RCA 15.007027A G.E. 25.007308 SYLVANIA 5.007581A SYLVANIA 15.00

PHONE44-208-684-1166

FAX44-208-684-3056

LANGREX SUPPLIES LTD

ECC81 RFT 3.00ECC82 RFT 6.00ECC83 RFT 8.00ECC83 EI 3.00ECC85 RFT 5.00ECC88 BRIMAR 6.00ECC88 MULLARD 10.00ECL82 MULLARD 5.00ECL86 TUNGSRAM 10.00EF86 USSR 5.00EF86 MULLARD 15.00EL34 EI 5.00EL37 MULLARD 30.00EL34G SOVTEK 5.00EL84 USSR 3.00EL519 EI 7.50EZ80 MULLARD 5.00EZ81 MULLARD 7.50GZ32 MULLARD 25.00GZ33/37 MULLARD 20.00PL509 MULLARD 10.00UCH81 MULLARD 3.00UCL82 MULLARD 2.00

A2900/CV6091 G.E.C. 17.50E82CC SIEMENS 7.50E83CC TESLA 7.50E88CC G. PIN TESLA 8.50E188CC MULLARD 20.00ECC81/6201 G.E. 5.00ECC81/CV4024 MULLARD 6.00ECC81/M8162 MULLARD 7.50ECC81/6201 G. PIN MULLARD 10.00ECC82/CV4003 MULLARD 10.00ECC82/M8136 MULLARD 12.50

SOCKETSB7G CHASSIS 0.60B9A CHASSIS 1.00OCTAL CHASSIS 1.00OCTAL MAZDA 2.00LOCTAL B8G CHASSIS 2.50

SCREENING CANSALL SIZES 1.00

STANDARD TYPES AMERICAN TYPES SPECIAL QUALITY TYPES

If you can find remnants of 1″ stock for the front panels withouthaving to purchase a full sheet, then the parts for the base mayeasily be adapted to be cut from the stock remaining in the ¾″sheet of MDF. Maintain the same outer dimensions for the base.

If you must buy a full sheet of 1″ MDF, the lumberyard will prob-ably do straight cuts for dividing the heavy panel to get it deliveredmore easily. I had mine cut at 39″ to transport it. If your lumber-yard attendant is really helpful, two cuts at 9″ and one at 13″ willdo most of your work for you.

If you must cut your own pieces, set the saw width to 9″ for thefront panels, then to 13″ for the top of the base and the front andback panels of the base. Reset for 3″ to cut off the base fronts and

backs. The base sides can be cut with the 3″ setting, then the fourbase fronts and backs cut from the end of the 13″ cut. Since thecrossover boards fit easily into the base, input connectors are bestlocated on the rear of the base. Rout out an appropriate amount ofthe rear wall to thickness which will accommodate your choice ofconnector hardware.

The cutting guide for the ¾″ sheet is also not very economical. Ifyour lumberyard cuts a 51″ piece from one end, you’ll probably beable to get the two pieces into your car. Set your saw for 12½″ forthe four sides, then to 7½″ for the backs and the T-line baffles. A 9″setting will provide the width for the tops.

The four corner pieces, which make up the “turns” for the cor-ners of the line, should have their long edges cut at 45° angles, 4¼″wide. Carefully trim one end of the baffle at an angle of 86°.

PLEASE NOTE:SEAS Fabrikker AS of Moss, Norway, has published a very earlyversion of the THOR cabinet drawing based on a preliminarysketch provided by author D’Appolito which has been on severalwebsites for many months. The drawing differs in a variety of dimensions based on 1″ MDF for the walls, top, and back panels. A new drawing is available on their web site http://www.seas.no/thor.html conforming to walls of ¾″ material. The dia-grams, cutting guides, and all dimensions published in audio-Xpress conform exactly to the D’Appolito prototype, with the ex-ception of the base where the prototype is made of ¾″ MDF.—E.T.D.

PARTS IDENTIFICATIONONE-INCH-THICK MDF

NO. PCS SIZE (IN.) PARTT Line Cabinet2 42 × 9 FrontBase2 5½ × 13 Top4 13 × 3 End4 13½ × 3 Side

THREE-QUARTER-INCH-THICK MDFT Line Cabinet4 41¼ × 12½ Side2 38½ × 7½ Back2 35¹¹⁄₁₆ × 7 Baffle2 12½ × 9 Top4 7½ × 4¼ Corners (Bevel 7½ side @ 45°)

THOR CUTTING GUIDE

“Reprinted, with permission, from audioXpress, Issue 5, 2002, p.p. 8-27, of audioXpress magazine. Copyright2002 by Audio Amateur Corporation. P. O. Box 876, Peterborough, NH 03458, USA. All rights reserved.”

Documents

THOR: A D’Appolito Transmission Linemadisound.com/pdf/Kits/audioXpress Thor Review.pdfVoice-coil diameter 39mm Suspension compliance 1.6mm/N Voice-coil height 16mm Suspension resistance