-HORNS REVISITED-

THE EXEMPLAR PROJECT
by Jeff Markwart and John Tucker
Triode Support Systems, Houston

Reprinted with permission from Sound Practices #9


 Almost Heaven

When we finished tweaking our Altec Voice of the Theater systems
as described in Sound Practices #4 we were happy campers. A mere
triode half-watt or so into those babies produced big smiles and
lots of foot tapping; movie soundtracks that sounded so right;
speed, dynamics! Life was good! 

Unfortunately, VOT life was not perfect. There are a few major
drawbacks that come with the territory when the home music
listener adopts a classic Altec auditorium design. To list a few:

*Size - A pair of A7s on spikes carrying 311-90 horns on top
visually dominates any normal size listening room. The decor
aspect of a VOT is best described as "Early Industrial Strength"
or "West Coast Monolithic".

*Bass - Not much below 40-50 Hz. A subwoofer or two is needed.
This sure doesn't help the size issue any.	

*Horn Integration - The aluminum exponential and the plywood
radius short horn just don't sound the same. 



After many experiments, we recognized that we stiffened, damped,
and tweaked the basic VOT technology well into the region of
diminishing returns. To successfully overcome the remaining
drawbacks meant major changes, a rethinking of the basic system
design to the point where the result probably wouldn't look or
sound much like an A7 when finished. 

We needed a smaller cabinet along with a lower system tuning
point to address size and bass issues. It turned out that Altec
had been building such a cabinet for many years, the Model 816.
The 816 is basically an A7 horn with a much smaller internal
reflex volume.

Well, we built a pair of 816 cabinets (ash veneer - still in the
shop somewhere) following Altec's plans, and stuffed them with
ALNICO 515Bs. Next came a matching pair of boxes to house the
811 horns, with provision for filling the interiors with sand.
We filled the exterior horn concavities with patching concrete
and mounted a pair of 806-A ALNICO compression drivers on the
resulting 811 module.

Nice looking system; small footprint, tuned at 40 Hz, crossover
at 800 Hz, 16 Ohm impedance. It sounded good in most respects
but the midrange was not as articulate as the A7. 

The reason turned out to be standing waves created between the
parallel non-flared sides of the 816 horn. With the sides at
16.125 inches the fundamental resonance was at 419 Hz. Altec
avoided this standing wave problem in the A7 bass horn with
non-parallel surfaces for the short horn sides. An additional
effect of the parallel sides in the 816 was a higher horn cutoff
due to a smaller mouth area compared to the A7, 177 Hz versus
160 Hz for the A7. These problems made a smooth transition out
of horn loading more difficult. 

Although our faith in the Altec drivers and horn technology was
high, we slowly realized that we would have to create a system
tailored to our goals to realize the full potential of horns in
a home listening situation. We wanted the entire system to have
a footprint smaller than an A7, offer a seamless blending of the
horns, and play deep, loud, and clear on a measly triode watt or
two.  

Enter the Tractrix 

It was during this period that we heard a full range speaker
system employing tractrix horn flares for the midrange and bass
drivers. The tractrix flare, although not new, has seen a
resurgence of interest in the last few years thanks to the
efforts of Dr. Bruce Edgar. 

The tractrix expansion is unique in that it is the only horn
flare whose mouth terminates at 90 degrees to its central axis.
It is also the only flare that can produce a spherical waveform
and it offers very low mouth impedance reflections compared to
other horns. The effect of this flare is a horn that does not
sound like a horn. The music is clean, clear and natural without
the characteristic horn signature. What a difference!  

We started to wonder:

Would high efficiency, professional quality drivers excel in
tractrix horns?

-Would compression drivers be suitable for use with tractrix
horns?

-Would a simple two-way design be feasible and adequate?

-Would our favorite vintage drivers be suitable for retrofit?

-Could bass reflex be effectively integrated with tractrix?


We decided that working out positive solutions to these
questions would bring us closer to our goal of overcoming the
A7's drawbacks and improving the overall listening quality of
our Altec-based horn systems. 


Bass Horn Module

We started off on our design quest with bass cabinet design. In
designing the bass horn we had to juggle a bewildering number of
interrelated factors. The going was slow as we wrestled to
produce the optimum blend of aesthetics, performance, and size
using spreadsheets and CAD tools.

A few of the major factors we dealt with were:

a. Horn shape - round, square or rectangular.

b. Mouth area/cutoff frequency/aspect ratio.

c. Overall height to allow acceptable compression driver horn
   mounting height.

d. Driver mounting depth to allow ease of compression driver
   time domain alignment.  

e. Optimum internal reflex volume versus
   the finished product are 33" tall, 26" wide, and 25" deep and
   the visual aspect is far more domestic than the old A7 bass
   cabinet (Figure 1). 

f. Materials and construction techniques

g. Tuning

Since direct radiation would be important in the region where it
blends with reflex energy, the bass driver would have to perform
from the deep bass through the upper midrange_a span of 6 to 7
octaves. We also needed high efficiency and a reasonable reflex
volume requirement. 

Our initial modeling efforts concentrated on finding a woofer
that met these requirements. Plot after plot of Thiele-Small
parameters yielded few contenders. Our favorite vintage Altecs,
416s and 515Bs were knocked out fairly early in the process due
to their low mass rolloff frequencies and large reflex volume
requirements. 

What we needed was a driver with a huge motor, a very light and
stiff cone assembly, and Fs/Qes/Vas values that allowed t

Unfortunately comments regarding our superdamped 511B horns and
902 compression drivers were also consistent - they were way
above average for 511Bs due to the sand damping but they didn't
really click. Nobody could put their finger on it, but something
wasn't quite right. These findings sure kept us motivated to
continue our HF tractrix horn development program!

One of the big lessons we learned during this period was how
different the horn profiles actually sounded from each other.
And when different types were mixed, such as a VOT bass with a



tractrix treble, or a tractrix bass with an exponential treble,
their unique signatures were very apparent, and didn't sound
well integrated. When we switched to a tractrix horn curve for
compression driver loading that problem went away and the LF and
HF sounded as one. The Exemplar system had been born. 

Tractrix Compression Driver Horn

The prototype tractrix compression driver horn was also
fashioned from wood, covered a narrow 35 degrees in the
horizontal, and had a mouth cutoff of 333 Hz. See Figure 3. It
was obvious from the first note that it shared the high
articulation and low distortion characteristics of the bass
module. 

Working prototypes were then produced with a wider 70 degree
horizontal dispersion, which more closely matched the bass
modules in coverage angle and sensitivity. The larger mouth area
produced a cutoff frequency of 232 Hz. We adopted this horn
profile for the final design as pictured throughout this article.

An interes
A composite magnitude plot of front horn/cone and rear port
radiation for the bass module is shown in Figure 6. 

The quasi-anechoic magnitude response curve for the compression
driver (Figure 7) reveals the basic characteristics they all
share; peaks centered around 2 and 4.5 kHz and falling high
frequency response. Interestingly, the tractrix did better in
the extreme top end than either exponential or constant
directivity horns, which were typically down by 10 dB and 20dB
respectively at 20 kHz. 

This peaked and rolled unequalized response is what gives
compression drivers their presence and speech intelligibility
and it also explains why a simple shelving control will not
produce adequate results in Hi-Fi applications. If you shelve
down the upper midrange to match the woofer the high frequencies
are lost; if you crank up the high frequencies the upper
midrange drives you over the edge. 

Figure 8 indicates what we accomplished with passive
equalization - same test setup, same driver. An additional
benefit to this equalization was that the compression driver
sensitivity could be matched to the woofer and thus eliminate
the need for series padding resistance in the high-pass circuit.
The resulting system sensitivity is 103 dB/SPL @ 1W/1m. 

A Better Crossover

The vast majority of two and three-way speakers are designed
with passive crossovers. They are relatively simple to implement
for the manufacturer and hold down cost and complexity for the
user. You just plug in an amplifier and play!

Unfortunately few commercial crossovers use premium parts and
bi-wiring. Even fewer can be easily bypassed to allow direct
access to the drivers and not many are contained in separate
outboard enclosures to eliminate the acoustic and magnetic
effects of a hard working woofer nearby.

The Exemplar passive crossover was designed with all of these
concerns in mind. Housed in a separate, sand-filled enclosure,
it is completely bi-wired from the supplied speaker cable right
up to the individual drivers. Reactive signal path components
were chosen for their sonic merit - Solo CFAC inductors and
Sidereal capacitors. 

For cable and point-to-point wiring we selected long crystal
copper developed for the Japanese super-conducting magnet
program. The crossover point and slope are 750Hz and 12dB per
octave, the filter type is Linkwitz-Riley. Internal enclosure
wiring is also super conducting magnet type, 12 AWG to the 
woofer and 18 AWG to the compression driver.

Drivers

Why do we use premium drivers anyway? After all, the Altecs cost
masny times more than the typical drivers found in most high-end
speakers. Wouldn't less expensive drivers have been just as 
good? Frankly, no.

Looks

The matching, complimentary, horn flares terminate at 90 degrees
and then continue to quickly curve away, and at the same time
into each other, via the rounded cabinet edges. Flat cabinet
frontal area is simply nonexistent, the horn is dedicated to
producing direct sound from the drivers. The curves lead the eye
inward as well along the junction of the surfaces, and give
visual form to the concept of the tractrix. 

Sound

At times the speakers utterly disappear in an acoustic sense,
like articles of furniture sitting mute amid sonic figures
arrayed on the soundstage. They can seem totally out of place. .
. we have found ourselves wondering "who parked these objects in
the center of all this musical activity?" 

So, after eight years of modifying, redesigning, substituting,
measuring, and countless hours of listening, we finally managed
to create a horn based system that does all the things we were
told horns aren't supposed to be able to do. Things like deep
and wide soundstaging from virtually any listening position in
front of the speakers; flat, extended treble response from a
compression driver, transparency; and emotional impact.

Obviously, we are very pleased with the results of our efforts.
We are also pleased to be part of the resurgence of horns and
triodes for home listening. By the way, some of the motivational
credit has to go to the "theys" who said it couldn't be done! 

Fig 1 _ Exemplar system shown with camo white A7 for comparison

Fig 2 _ Experimental d'Appolito configuration using bass horn
        prototypes with sand-damped Altec 511B  HF horns 

Fig 3 _ Early prototype 333 Hz 

Fig. 7_  Unequalized system response

Fig.8 _ System response with passive EQ

Interior aspect of Exemplar cabinets showing ducted port
construction.