Summary
The Eros project uses the Vifa PL18 6.5" woofer and Scan Speak 9500 tweeter
in a D'Appolito-style MTM configuration to produce a high-quality speaker
with good sensitivity, excellent dynamics, and very musical, full-range
response.
Specifications
Response: 37hz-20khz (-3db)
Impedance: 4 ohms
Sensitivity: 90.5 db
Designer Note
An excellent speaker that produces a wide, deep soundstage and pinpoint
imaging. Recommended amplifier power is at least 10W, but due to its
impedance profile, this project is recommended for use with amplifiers that
can handle low impedance speakers.
Construction Photos
|
Eros!
Eventually, we all
want to take the jump into a "high-end" speaker. When you go to a retail store
or search around the web, it becomes apparent that to take this jump, you
either have to spend alot of money or you have to set your sights a little
lower. Those of you who have been willing to put a little time and energy into
building your own speakers know that for a mere fraction of the retail (or
even deeply discounted) cost of commercial speakers, you can build a set of
truly audiophile quality speakers yourself.
There are numerous excellent sounding speakers sold in "kit"
form by various vendors. These kits will provide you a level of performance
worthy of the term "audiophile". Many of these kits, however, are still beyond
the reach of the average DIYer. Moreover, some of these high-end kits can cost
over $1500 and still not have wrung every last bit of performance out of the
drivers and crossover network -- to me, good engineering is worth alot more
than expensive drivers, so a less expensive speaker that's better designed
will sound better than a more expensive speaker that doesn't live up to its
full potential. So, just as you all know that I firmly believe that speakers
don't need to cost alot to sound great, I believed that I could make a
top-quality speaker that compared well with anything else out there without
breaking the bank. And this began the VSS Eros Project.
I. The Drivers
As with every speaker design,
the first step is selecting the best drivers for the job. In this case, I
selected the Vifa
PL18 woofer because I had heard it produced the exceptional, highly
regarded midrange clarity of the Vifa
P17WJ but with the deep, highly resolved bass of the Scan
Speak 8545k (without the high price tag of the Scan Speak woofer). From
my experience with the Dayton
III design, it became apparent to me that two woofers were going to be
better than one for a more efficient and highly dynamic speaker.
|
 |
For the tweeter, I thought
about a wide range of units. My belief is that the more you have to
attenuate the tweeter to match the level of the woofers, the more you will
compress the sound coming out of the tweeter and the less dynamic it will
sound. This lead me to look at tweeters with a sensitivity in the range of
89-91db. Second, I wanted a tweeter with excellent, open sound capable of
very detailed reproduction but one that used ferrofluid in the air gap for a
low resonant frequency and good power handling to accommodate a low
crossover frequency without the need for an elaborate, expensive conjugate
filter to reduce to the impedance peak at resonance. This narrowed the
selection down to units from Scan Speak, Seas, and Morel. From among these
manufacturers, I chose the Scan
Speak 2905/9500 because it has a reputation from being one of the best
sounding drivers that met these criteria. Also, I felt that this driver
would have the best sound with a crossover frequency under 2khz that I
wanted to target to maximize the off-axis performance of the woofers.
Finally, my impression of Scan Speak drivers is that they have a lively,
articulate quality without even a hint of harshness.
II. The Enclosure
My preference for this
project was a floorstanding enclosure. With that being the only caveat, I
then measured the thiele-small parameters for the woofers after a thorough
break-in period.
|
|
|
| The
PL18's tested almost identically (a good sign). |
I
then measured the frequency response of the woofers, splicing the
ground-plane measurement together with the farfield response at about
200hz. |
I then designed the
enclosures. I selected an SBB4 alignment tuned to approximately 38 hz to
give the best combination of low-end response coupled with a shallow
(18db/octave) rolloff. This produces the best transient response in a ported
design and lowest group delay for the cleanest, best resolved bass possible
without using a sealed or transmission line enclosure.
The resulting enclosure has a
net internal volume of 1.64 ft^3 and a vent port that is 3" (id) by 3
3/8". While this value is not exactly as predicted by the standard
equations, the port was tuned through trial and error. After building the
enclosures, I tested the impedance response and trimmed the vent tube until
the exact resonant frequency was achieved. The bass response of the final
enclosure is simply exceptional -- very deep and clear with wonderful tonality.
A very important factor for achieving this level of performance was the
construction of an extremely solid enclosure. The front baffle is 1
1/2" thick mdf (made by gluing together two pieces of 3/4" mdf and
clamping them overnight prior to cutting the piece to the appropriate size.)
All of the internal surfaces were covered with 1/2" birch plywood for
added ridgidity. The front baffle is 9.5" wide with an outer edge
rounded with a 1 1/2" roundover router bit and the inner edge rounded
with a 3/4" bit. This was done in combination with offsetting the
woofers drivers 5/8" to the inside of the baffle's vertical centerline
to improve imaging. The tweeter is offset 1 1/8" from the centerline
toward the inner edge of the front baffle. The center to center distance
between each woofer and the tweeter is 6.75" for reasons that will be
more fully explained in the crossover design section. Don't be surprised
that this number is somewhat larger than the center to center distance
recommended by other designers, as my experience has found that this
approach provides a taller, deeper soundstage with less compression in front
of the speaker at the expensive of a slight amount of imaging that can be
regained with a good quality amplifier and proper speaker placement in the
listening room.
If you have questions
regarding the construction of the enclosures, feel free to email
me. If you wish to build the Eros speakers,
please
download these high-resolution mechanical drawings of the enclosures. (2.0
mb)
III. The Crossover
The heart and soul of any
multi-way speaker is in the crossover design. To develop the crossover, I used
the Clio measurement system to obtain the frequency response and impedance of
the drivers while mounted in the actual enclosures. I then imported the
measurements in Calsod Pro 3.10g to model and simulate the crossover. Many
different crossover topologies were tried, including some that produced
exceptionally flat measured response, as that seen below.
|
|
 |
|
|
Flat didn't
necessarily translate into the level of performance I was looking for
and also resulted, in some instances, in complicated crossover networks
with many components. I believe strongly that the number of components
in the signal path must be kept to a minimum to allow the drivers to
reach their maximum dynamic potential and to sound as open and natural
as possible. This eventually lead me to the following crossover network
in which only a signal component is in the signal path to each driver.
In fact, even the attenuating resistor for the tweeter was moved in
parallel to the driver (thanks for computer optimization) and final
tweaking was done by ear to provide a pristinely accurate, yet very open
and natural sound. The final result is what I think is an outstanding
speaker with high efficiency and very lifelike dynamics and sound
quality. I'm currently using these speakers with my monoblock 300B
amplifiers that produce 8 watts/channel and they are capable of
incredible output. |
Please note that
the woofers are connected in parallel and all drivers are connected with
normal polarity. The crossover frequency of the final version is around
1750hz and the woofers are rolled off with a 2nd order acoustic slope.
The tweeters also use a 4th order acoustic slope. The relatively low
crossover frequency permits a wide center to center driver spacing to
improve the quality of the soundstage without succumbing to the
traditional compression of the soundstage that many MTM designs are
plagued with. The mixed-order slopes provide better phase tracking of
the drivers to compensate for the offset of their acoustic centers. As a
result of the excellent phase tracking, the drivers sound exceptionally
well integrated, as is evidenced by a 38db reverse phase null (not
shown). |
 |
|
|
The overall response is flat to
+/-<3db from about 40hz to 20khz. The peak from about 600hz to about
1khz is inherent in the woofers used and suppressing the peak did not
improve the sonic quality of the speakers. The impedance is a nominal 4
ohms with a minimum of 2.75ohms at around 5000 hz. While the impedance
minimum may be a slight problem for low-quality amplifiers, no audible
distortion is heard with my 300b amps, nor with various solid state
receivers. Due to the very benign impedance phase angle (less than +/-
20 degrees), even moderately powered amps can produce very dynamic sound
with these speakers. Those with significant impedance concerns
should consider the Eros MKII project. |
IV. Listening
Impressions and Final Thoughts
Without hearing these speakers, it's
almost impossible to describe why I feel so strongly about them. I've heard
many speakers ranging in price from $20 to $20,000 and these speakers have
an overall magic to their sound quality that would let them sit proudly
among even the most highly acclaimed commercial speakers. The imaging and
openness of these speakers is stunning at times, and the dynamics are truly
sensations. The drivers are capable of excellent resolution with low
distortion to cleanly resolve complicated musical passages while retaining
delicate timbres.
View
Parts Express Parts List for This Project
(this list is only current as of the time of the original posting of this
project)
|
