You Can DIY! - Circuit Cellar

Transcription

INNOVATIONS IN AUDIO • AUDIO ELECTRONICS • THE BEST IN DIY AUDIO
audioXpress. Reprinted by permission. For subscription information, call 800.269.6301, or visit www.audioxpress.com. Entire contents copyright © Segment LLC. All rights reserved.
www.audioxpress.com
www.audioxpress.com
audio xpress
ADVANCING THE EVOLUTION OF AUDIO TECHNOLOGY
FAST Company
QSC’s Flexible Amplifier Summing Technology
You Can DIY!
Speaker Rebuild
with Heil
Transformer
Resurrect Classic
Speakers or Design a
New System
with DSP
Sound Control
We Really Have (or Want)
Great Acoustics
Richard Honeycutt’s New Column
Speakers
An Introduction to Woofers
Welcome to the Low-Frequency World
Practical Test & Measurement
Designing for
Ultra-Low THD+N
NOVEMBER 2013
ax
You Can DIY!
By
Thomas
Perazella
Use a Heil Transformer and an
Updated Woofer
T
(United States)
Tips to Resurrect a
Classic Speaker or Design
a New System (Part 1)
IN
This article is the first of a three-part series that will describe
the history, construction, testing, problem solving, and
voicing of an old speaker brought back to life by new
technology. Although the original goal was to breathe
new life into a classic, many of the issues addressed
EP
R
here are also critical when designing a new speaker or
even getting the best performance from commercial
speakers and associated equipment.
But there is also a lot of misinformation on how it
really works. It has been said that the diaphragm,
which is a vertically aligned array of pleated material behaves like an accordion squeezing the air in
and out. Actually, the whole diaphragm does not
move in and out like an accordion, rather alternate
folds move either in or out so the air is squeezed
out one side and sucked in the other. The folds
stop the diaphragm from moving back and forth
so the result is a dipole pressure wave. For a better understanding of this device, please refer to the
Resources section.
What is an AMT?
Heil AMT History
R
Photo 1: The Eton 7” drive
has a light rigid cone made
of a Nomex honeycomb
sandwiched between two
Kevlar layers for true
pistonic behavior.
When it comes
to musical taste,
you could probably classif y me as an
omnivore. Depending on my
mood, I may lean toward classical,
jazz, rock, pop, country, or other genres. Like most
of us, there is one special piece that goes directly
into my heart, bypassing all logic to create an
intense emotional involvement. For me, it is Gustav
Mahler’s Symphony No. 2, “The Resurrection.” It is
also a fitting title for my latest completed project.
Several years ago, I became interested in the
Heil Air Motion Transformer technology. Created
by Dr. Oskar Heil, this device uses an unorthodox
mechanism to quickly accelerate air, resulting in
a very interesting tweeter. There is a lot written
about the actual mechanism, including several patents leading up to the AMT version (see Resources).
The AMT was originally produced by the California company ESS, but the final complete speaker
received mixed reviews. It was a two-way design
that suffered from running the AMT at too low a
frequency and the 12” woofer at too high a frequency. The two drivers’ radiation patterns were
different because of their physical shapes and
46 | November 2013 | audioxpress.com
T
IN
Photo 2: The cavities in the Heil AMT housing were filled with deadening material in an
attempt to reduce cavity resonances.
EP
R
the wavelengths reproduced vs. the size of each
throughout their crossover frequency ranges. Also,
one was used in open air as a dipole and the other
in a box as a monopole. The AMT’s crossover point
and slopes caused it to operate at the low end in a
range that did not work with the given slopes. The
resulting speaker did not live up to the AMT’s true
performance capabilities. In spite of this, people
loved the speaker so much it almost became a cult
classic because of the AMT’s crystalline response.
I do not remember exactly when I decided to
use the AMT as the basis of a DIY project, but it
was probably in the late 1980s or early 1990s. I
obtained a used pair of the tweeters and immediately knew I would have to use a three-way
configuration if I was going to be able to get a
wide enough dynamic range without exposing the
tweeter to undue stress. I would also have a chance
to better control the integration of the drivers’
radiation patterns.
Although a three-driver three-way is a common configuration, some severe compromises in
a one-driver-per-frequency-range configuration
must be addressed. Primarily, they have to do
with sensitivity, linear displacement capability, and
power requirements, which I recalled during the
implementation. It also reminded me why I moved
on to other solutions. However, the concept has
validity because of its smaller size and lower cost.
My Original Project
R
The speaker concept used a sealed box holding
a dynamic driver for the bass, a dipole mounted
dynamic driver for the midrange, and the AMT
as a dipole for the high frequencies. I located
a woofer box and grille used in an exceptionally
well-performing speaker, the Acarian Alon IV. It
was designed for a sealed-box woofer and a dipolemounted midrange and tweeter (see Resources for
more information). I then had to choose the bass
and midrange drivers.
When I designed the original speaker, the Ger
German company Eton had developed a woofer/mid
woofer/midrange driver that used what it called “Hexacone”
technology. The cone was made from a Nomex
honeycomb sandwiched between two Kevlar layers
(see Photo 1). The result was a cone that was very
light but extremely rigid. Within a certain range,
it acted like the ideal piston.
In addition to the sandwich cone, the Eton
woofer/driver had an inverted soft rubber surround. The result was a relatively smooth, low-distortion output, even when driven hard. There were
some ripples in the passband and if you went up in
frequency, an anomaly in the frequency response
Photo 3: The baffle housing
the midrange and tweeter
is isolated from the bass
cabinet by a rubber sheet
and rubber tubing in the
holes that receive the
indexing pins.
would appear. I have been told that the ragged
difresponse at higher frequencies was due to a dif
ference in impedance between the rigid cone and
the compliant surround that resulted in reflections
frequenof energy back into the cone at the higher frequen
cies. I cannot verify the actual mechanism myself,
but response problems do appear.
The solution to using a driver that has great
performance in most of its range but problems
at frequency extremes is simple. Don’t use the
driver in the problem areas. That may be problematic when using passive crossovers, but with
the flexible electronic crossovers now available,
it has become a non-issue. At the time, I decided
to use the Eton woofer/driver with a home-built
12-dB/octave electronic crossover. I mounted it
in a quasi-dipole configuration with an open back
but heavy damping. As it turns out, that was okay,
but not the best solution.
The original AMT driver was mounted in a plastic housing that had a ridge around the front and
back that created triangular-shaped cavities at
the top and bottom in the front and rear of the
radiating element. Although I had no way to measure those cavities’ effects, I altered them by filling the cavities with wool carpet underlayment
audioxpress.com | November 2013 | 47
ax
You Can DIY!
material. In addition, I fastened some thin black
felt material over the faces of the pole pieces to
improve the tweeter’s looks and possibly reduce
some reflected energy (see Photo 2). It is interesting to note that the AMT’s current version has
the cavities removed from the housing and has
added what appears to be a black surface-finish
treatment on the pole pieces.
Both the AMT and the Eton drivers were
mounted on a flat baffle that sat on top of the
woofer box in positions to time align the three drivers. I added mass to the baffle base in the form of
a box containing cement. I used sand-filled steel
tubes to brace the baffle’s upright section to the
base. The baffle was located on the woofer housing with steel pins fastened to the housing and
slip fitted into holes in the baffle’s base. I used
surgical rubber tubing inserts to isolate the base
from any vibration that may have come up the
pins. The baffle’s base also had a rubber sheet on
the bottom for isolation. Photo 3 shows the layout
of the baffle’s base.
For the bass range, I originally used 12” drivdrivers that I already had. However, they did not have
much linear excursion, so I switched to 10” PeerPeerless drivers. They were better, but the additional
excursion hardly made up for the loss in cone area.
Ultimately I used 12” drivers from HSU Research
that had a higher X MAX and a larger cone area.
Photo 4 shows the completed speaker.
I used those speakers for many years with good
results. However, as time went on, I realized they
had limitations, including limited bass excursion
and a radiation pattern that resulted in more floor
and ceiling reflections than I preferred. Ultimately,
I replaced them with my current main system that
uses Bohlender-Graebener RD75 drivers for the
high frequencies, Peerless 831727 10” drivers for
the upper bass/lower midrange, and sealed box
subwoofers. (More information about these drivers
and midranges is available in my article series,
“ On Angels Wings,” Part 1 and 2, audioXpress,
January–February, 2001 and “A Dipole Midbass,”
Part 1 and 2, audioXpress, June–July 2004.) The
Heil-based speakers were put into storage. With the
AMT tweeters available to the public once again,
I decided to see if I could improve on the origiorigi
nal design, especially since a range of powerful
reasonably priced digital signal processing (DSP)
equipment is available.
Updating the Design
T
Photo 4: The original
speaker used an HSU 12”
bass driver, which is shown
without the grille.
R
EP
R
IN
Although the AMT and Eton drivers have under
undergone revisions over the years, both versions I had
are still adequate for the job so I did not replace
them. Woofer designs, however, have progressed
dramatically in the intervening time. The search
was on to find a suitable substitute.
A major advantage of current speaker control
devices is that trying to achieve the woofer’s critical
tuning in an enclosure is not as much a necessity
as in the past. For example, if you use a passive
crossover with an incorrect speaker parameters
and box volume combination, you may wind up
with a Qb that is too high. The result would be
a bump in the frequency response at resonance.
In the past, that could result in an overly bassheavy voicing. Most DSP-based electronic crossovers enable you to create a bandpass filter with
any center frequency, Q, and any amplitude you
desire to flatten out the response bump. Because
of this, if you already have an enclosure as I did
or are restricted by severe space limitations, your
driver choices are greatly expanded. You do not
have to worry about achieving critical damping by
choosing from a limited range of drivers with the
correct parameters for the given box size.
Photo 5: The Dayton
Ultimax UM12-22 woofer
has several great features,
including 19-mm XMAX.
The Woofer
When it comes to speakers with drivers covering separate frequency ranges, the woofer is the
group’s heavy lifter. For each octave the reproduced frequency decreases, four times the volume
displacement of air is necessary to maintain the
same sound pressure level (SPL). If you need “X”
amount of volume displacement to achieve your
desired SPL at 20 kHz, at 20 Hz you need more
than 4 million times “X” to achieve the same SPL.
Fortunately, music does not produce the same SPLs
at the highest frequencies as in the bass, but it
quickly becomes apparent why so few speakers
can reproduce high-amplitude organ pedal notes
at realistic listening levels without excessive distortion, if even at all.
a)
T
Photo 6: A pivot plate was constructed and inserted into the original woofer mounting
hole to provide the alignment point necessary for the Jasper jig and router to properly
enlarge the hole (a). The enlarged hole in the bass cabinet is shown with the backer plate
and pivot hole before routing of the lower basket clearance hole (b).
Mounting the Woofers
IN
After receiving the new woofers, I removed the
old ones from the enclosures to see what modificamodifications were needed to fit the new drivers. The new
drivers turned out to be 0.25” larger in outside
diameter than the old ones, so I had to enlarge
the original mounting holes to accept the baskets.
In addition, the mounting flanges and surrounds
were a bit thicker. To provide clearance for the
grilles and give it a more finished look, I recessed
the drivers further into the housing.
This combination presented two challenges.
The first was that I no longer had a center hole
to use with my Jasper jig and router to enlarge
the hole. The second was that, even though the
enclosure’s front panel was 1” thick, the amount
of recess I desired would leave too little material
after routing to securely fasten the drivers.
To solve these problems, I decided to use a
enclobacking plate mounted to the inside of the enclo
sures. First, I cut out two pieces of 0.75” MDF to
mount inside the enclosures. The pieces covered
the existing holes and extended far enough to provide the necessary strength to mount the drivers.
They also acted as a support for the router I used
to enlarge the hole’s diameter. Two pieces were
cut to fit inside the enclosure at the front.
A small problem cropped up when I tried to
insert the backers. The enclosure’s walls had additional 0.5” thick MDF plates with deadening material fastened to the inside to reduce vibrations.
Because of the backers’ size, the edges of the
plates attached to the sides interfered with the
backers. I removed, trimmed, and set them aside
to be refastened inside the housing after all the
work on the backers was completed. Then, I temporarily screwed the backers into place against
the front plate.
Next, I fabricated a temporary pivot plate to
fill the existing hole flush with the enclosure front.
R
EP
R
To achieve the high linear volume displacements
necessary for clean bass, you need a lot of radiating surface combined with a lot of linear excursion. It is similar to an internal combustion engine
where displacement is a product of the cylinder
bore times the piston stroke times the number of
cylinders. If the engine needs more displacement,
you must have some combination of more pistons,
bigger pistons, or a longer stroke for the pistons.
In this speaker’s case, the “bore” was limited
to a 12” driver and the number of “cylinders” or
woofers in each speaker was one. To get the volume up to a reasonable level, a high linear “stroke”
or excursion was needed.
The specification for a dynamic driver’s maximum linear one-way excursion is X MAX . It is the
excursion where either the motor force factor (Bl)
falls to 50% of its peak value or the suspension
stiffness (KMS) increases to double the initial value.
X MAX is defined as the direction of travel either in
or out that has the worst of either of the limiting values.
For example, if a driver is limited to 5-mm
travel in one direction by force factor reduction
and is limited in the other direction by increased
stiffness at 4 mm, the driver’s X MAX will be 4 mm.
Unfortunately, some manufacturers quote maximum total excursion in both directions regardless
of either parameter’s nonlinearity as the excursion
instead of true X MAX . Caveat emptor!
In previous woofer designs, I used drivers
from various companies that had different linlin ear excursion vs. price combinations. Some have
incredibly
credibly high linear volume displacements in
small drivers but at a high price. Over the years,
my best results for drivers with high linear disdisplacement at reasonable prices have come from
Dayton
yton Audio, the house brand of Parts Express.
My current reference system uses two Dayton 15”
DVC woofers that provide yeoman service for a
two-driver arrangement.
Looking at the Parts Express website, the sub
subwoofer section brought me to a relatively new
series of drivers called “Ultimax.” This series con
consists of three drivers in 10”, 12”, and 15” diame
diameters. They all have similar features, which can be
viewed on the website. One with a 19-mm X MAX
(which is huge for a driver in this price range)
was really impressive. I decided to use the Dayton Audio UM12-22 12” driver. It was not so many
years ago that drivers with these specifications
would have been unheard of, let alone affordably
priced. Anyone who refers to “the good old days”
obviously didn’t build speakers. I purchased two
units (see Photo 5).
b)
About the Author
Thomas Perazella is a
retired IT director. He
received a BS from the
University of California,
Berkeley campus. He is
a member of the Audio
Engineering Society, the
Boston Audio Society,
and the DC Audio DIY
group. He has written for
Speaker Builder and audioXpress magazines. He
has authored several articles in professional audio journals and taught
commercial lighting at
the Winona School of
Photography.
audioxpress.com | November 2013 | 49
ax
You Can DIY!
screws and I drilled not only through the plate but
on through the backer to make a pilot hole for the
jig. This assured me that the enclosure’s newly
routed outside clearance hole would be exactly in
line with the inside driver frame clearance hole
that I later routed (see Photo 6a).
Then, I set the proper spacing on the Jasper jig
and routed the new mounting hole in the encloenclo
sure. Once the hole was finished, I removed the
pivot plate and checked its fit with the driver.
Note that the backer plate has the four holes that
were used to fasten the pivot plate and also the
pivot hole that was then used to route the clearclear
ance hole in the backer. Because the new driver
needed extra depth, the additional routing opened
two small holes into the original enclosure’s inside
(see Photo 6b).
6b). Although they would be covered
by the driver frame and gasket, I decided to use
the wood filler to ensure the enclosure was sealed.
Next, I removed the backers and routed the
clearance holes through which the rear of the driver
baskets would fit. When that was done, I glued
the backers and screwed them to the enclosure’s
inside using the same holes to ensure alignment.
I replaced the damping plates. Then I masked,
primed, and painted the front of the enclosure.
To maximize the enclosure’s apparent inside
volume, I stuffed it with the fiber material that had
been used in the original speaker. Next I wired the
drivers using 12-gauge zip cord and a single piece
of 12-gauge wire as a jumper between the two
voice coils. To achieve 4-Ω impedance, the two 2-Ω
coils on each driver were wired in series with one
coil’s negative terminal wired to the other’s positive terminal. I wired the remaining positive and
negative terminals from each of the two coils to the
input terminals on the enclosure (see Photo 7). I
used McFeely’s #8 black-oxide square drive wood
screws to fasten the two woofers. I have been
using square drive screws in critical applications
for years as they eliminate the “camming” that
often happens with Phillips screws. McFeely’s has
a good selection. The results provided a secure
mounting for the woofers and, with the additional
depth to the recess, a clean look. Next I moved on
to the midrange and treble sections.
I will cover the remainder of the construction
phase plus some problem solving in Part 2 of this
article series. I will continue with problem solving
plus voicing and music correction curves in the
final article in this series. ax
T
Photo 7: The woofer
terminals were connected
to the terminal plate using
12-gauge zip cord. A
12-gauge jumper was used
to tie the two voice coils
together in series.
EP
R
Resources
IN
This plate enabled the router to smoothly slide over
the enclosure and provide the necessary pivot hole
for the Jasper jig. I used a scrap piece of plywood
for the plate that was not quite thick enough to
sevpresent a totally flush surface by itself. I used several temporary shims of the proper thickness on
top of the backer to solve the problem. I screwed
the plate to the backer with four flat head wood
R. Deutsch, “Acarian Alón IV loudspeaker,” 2006, www.stereophile.com/floor
loudspeakers/293acarian/index.html.
Eton, “Home Hi-Fi,” www.etongmbh.de/en/products/home-hifi/midrange-bassmidrange/7-360-37-hex/1/pid/163.
Parts Express, www.partsexpress.com.
T. Perazella, “On Angels Wings, Part 1,” audioXpress,
audioXpress, January 2001.
———, “On Angels Wings, Part 2,” audioXpress
audioXpress,, February 2001.
———, “A Dipole Midbass, Part 1,” audioXpress
audioXpress,, June 2004.
———, “A Dipole Midbass, Part 2,” audioXpress
audioXpress,, July 2004.
US Patent and Trademark Office, Oskar Heil patents, www.google.com/patents/
US3636278.
R
———, Oskar Heil patents, www.google.com/patents/US3832499.
Wikepedia,, “Air Motion Transformer,” http://en.wikipedia.org/wiki/Air_Motion_
Transformer.
———, Oskar Heil, http://en.wikipedia.org/wiki/Oskar_Heil.
Sources
Ultra Drive Pro DCX2496 crossover and Ultra Curve Pro DEQ2496 equalizer
Behringer | www.behringer.com
UM12-22 Ultimax DVC subwoofer
Dayton Audio | www.daytonaudio.com
Heil Air-Motion Transformer tweeter
ESS Laboratories | www.essspeakersusa.com
5-880/25 Hex midrange
Eton GMBH | www.etongmbh.de
www.audioxpress.com
www.audioxpress.com
audio xpress
AUDIOXPRE SS | DECEMBER 2013
Acoustical Absorption
The Oldest Tool in the Modern
Acoustician’s Tool Box
You Can DIY!
The Cathedrals Speakers
By Ken Bird
elysia xfilter 500
Surface-Mount Magic in a Small Format
Audio Electronics
Build a Sound Level Meter
and Spectrum Analyzer
By Ron Tipton
Standards Review
From Audiobus to
Inter-App Audio
The New Mobile DAWs?
DECEMBER 2013
ax
You Can DIY!
By
Thomas
Perazella
A Viable Solution to Speaker
Sensitivity Problems
T
(United States)
Tips to Resurrect a
IN
This is the second installment of a three-part series about resur
resurrecting an old speaker that used some interesting technology. This
article will describe the speaker’s final construction. I will also offer
a solution to a sensitivity issue that impacts not only this design,
R
EP
R
but possibly the speakers you are currently using.
Photo 1: The completed
speaker is shown from an
oblique angle with the grille
in place.
Photo 1 shows the revamped
speaker, which closely resembles
the original design. In the orig
original speaker, the midrange and
tweeter were mounted on an
open vertical baffle that tapered
from bottom to top and had
rounded edges. That was done
to minimize diffraction. At first
I thought this section could be
used without changes, but that
was not the case.
The Top End
T h e H eil t we e t e r s we re
mounted to a shelf on the baffle
using the original bolts that came
with the drivers. When doing some
preliminary sine wave sweep testing,
I noticed the housing on the tweeters
would resonate at approximately 160 Hz.
Vibrations generated by the woofers traveled from
the enclosure through the baffle even though a
rubber pad separated the enclosure from the baffle and the locating pins were isolated by pieces of
latex rubber tubing.
To solve the vibration problem, I resorted to an
isolation technique I had successfully used with shop
machinery. Sorbothane rubber is one of the best
vibration absorbing materials. To achieve effective
vibration isolation remove as much of the vibrational energy as possible from the system. Don’t
try to just re-transmit at a different frequency or
amplitude combination. Sorbothane is effective at
converting vibrational energy to heat. Depending
on the durometer the first time you use it, the
consistency may seem like a gum rubber eraser
or even a squishy rubber version of well-chewed
Turkish taffy. For this application, I chose a 0.25”
sheet of 30 durometer, which definitely falls into
the Turkish taffy arena.
My original question was how to achieve the
best isolation from the lower enclosure. Instead of
using the Sorbothane at the baffle’s base, I decided
to isolate the Heil from the baffle. This made the
isolation effective against potential woofer and midrange induced vibrations. I decided to use Sorbothane sheets to sandwich the air motion transformers (AMTs) on the top and bottom. I held them
in place by making plates with attached threaded
50 | December 2013 | audioxpress.com
Photo 3: Surgical rubber
tubing cut into appropriate
length pieces is an effective
method to isolate vibration
between the baffle’s support
shelf and the mounting
bolts.
R
EP
R
When designing a speaker, as with almost
everything else in life, the key to success is making informed compromises. If audio is your hobby,
your efforts should make you happy, or why bother?
Don’t let anyone else tell you the “correct” way to
do things. As former-President Ronald Regan once
said, “Trust but verify.” When it comes to audio,
you need to know what you expect from your music
reproduction.
That being said, before voicing or designing a
speaker, you should define the reproduced sound’s
most important characteristics. Then try to make
the fewest number of compromises in the areas
most important to you. For me, I define them in
order of preference—flat frequency response, high
dynamic range, broad frequency response, concon trolled radiation pattern, good transient response,
and reasonably low distortion.
These criteria should not be viewed as standalone parameters as they definitely can and do interinteract with each other. For example, if your speaker
does not have enough linear volume displacement
prowhen you try to achieve the output levels that pro
duce high dynamic range, you will receive high
distortion levels.
Regardless of your preferences, in a multidriver speaker you should first decide what kind
of crossover you want to use—passive or active.
Next, you should define the crossover points and
slopes because they will affect other considerations.
Then, determine each driver’s drive level. You should
also address any driver performance anomalies.
In my case, the first decision was a slam dunk. I
have not used passive crossovers for years. Designing a comprehensive crossover is more difficult
using passive components and inevitably results in
too many compromises. Also, when working with
passive components, any errors that creep into
your initial assumptions require a lot of time and
T
Crossover Issues
Photo 2: To secure the
AMT tweeters to the baffle
assemblies, I made brackets
from a steel plate with a
0.25” × 20” steel-threaded
rod brazed into holes.
IN
rods to clamp them to the platform on the baffle
(see Photo 2). I placed the rods from the clamping plates through holes drilled in the platform. I
isolated them with the same rubber tubing I used
with the baffle locating pins from the baffle base
(see Photo 3). I also isolated the nuts and washers used to fasten the plates using small wooden
backers with Sorbothane pieces to separate them
from the platform (see Photo 4). The entire tweeter/
mounting plate assembly was effectively isolated at
every contact point. This completed the project’s
construction phase. The finished speaker with and
without grilles is shown in Photo 1 and Photo 5,
respectively.
Photo 4: Mount the tweeter on the baffle shelf with layers of Sorbothane between the
tweeter and rubber tubing around the bolts. Also use Sorbothane under the bolt mounting
plates.
audioxpress.com | December 2013 | 51
You Can DIY!
money to change parameters. Many options
for electronic crossovers provide several
choices of crossover types, frequencies,
and slopes at a low cost. Some of them
enable you to adjust for driver offsets
through time delays, set up base parametric EQs, and implement frequency
sensitive dynamic EQs that can help
mitigate sensitivity-driven amplifier
distortions.
Some even enable you to link
a PC and make changes, saving
virtually unlimited numbers of
combinations, while listening
to the music. Just the ability to
hear the effects as you make
changes is reason enough to
use electronic crossovers. If
you have not designed a speaker
with electronic crossovers, you
owe it to yourself to give it a try.
I predict you will never go back
to passive crossovers.
I used a Behringer ULTRADRIVE PRO DCX2496 digital
loudspeaker management syssystem for the crossovers. I have
used this device for years in
my main system. It provides a
degree of flexibility and control
that is hard to believe. There are
an incredible number of parameparame
ter choices when you design and
sysuse a speaker. In my main sys
tem, I used Linkwitz-Riley configu
configurations with 48-dB/octave slopes. I
used this configuration as the starting
point for this project. My original choices
for crossover points were 151 Hz and 1 kHz.
Later testing required changes, but that took only
a few seconds with this crossover.
low-frequency reproduction. The inevitable result
is a reduction in sensitivity, which is common with
most speakers today.
Because my design is a tri-amped speaker, I chose
three amplifiers I already had available. I used a Crown
Audio Studio Reference amplifier rated at 1,160 W per
channel into 4 Ω for the bass, a Hafler DH500 power
amplifier rated at 255 W per channel into 8 Ω for the
midrange, and a Parasound HCA 800-II amplifier rated
at 150 W per channel into 4 Ω for the high frequencies.
Considering the three drivers’ sensitivities, the
amplifier choices were clear. The Dayton Audio
UM12-22 Ultimax DVC subwoofer sensitivity was
84 dB with 4-Ω impedance. The Eton midrange
driver sensitivity was 89 dB with 8-Ω impedance.
The AMT tweeter driver sensitivity was 92 dB with
4-Ω impedance.
Beware of another misleading trend that is
presoccurring: Sensitivity is given as the sound pres
prosure level (SPL) at a 1-m distance created by pro
viding the driver with 1 W of power. For a driver
with 8-Ω impedance, that represents a 2.83-VRMS
voltage. Many specifications are now given with a
imped2.83-V drive level regardless of the driver imped
repance. If you do the math, that voltage level rep
resents 2 W with a 4-Ω impedance and 8 W with a
2-Ω impedance. That makes low-impedance drivers
look much more sensitive than their higher-impedance counterparts. However, if they are driven with
the same voltage, they are actually consuming more
power. The reasoning for this logic is the ability of
some voltage-limited amplifiers to provide more
power into lower impedance loads. However, they
still require more electrical power to achieve their
stated acoustical output.
If you take into account the drivers’ sensitivity,
Table 1 shows the driver/amplifier combinations
are good matches as far as maximum SPLs are
concerned. If all the peak levels at every frequency
in the music are the same, the amplifiers shouldn’t
limit dynamic range. As it turns out, that is a bad
assumption because of the high levels in the bass
range with some well-recorded material. On drum
recordings, I actually clipped the Studio Reference
amplifier. It was interesting to note that this powerful amplifier clipped before the Ultimax driver
complained, which is a true testimonial to this driver’s exceptional performance. I will discuss a rather
elegant solution to this problem later in the article.
Another problem occurred when I set the amplifier gains. The DH500 amplifier had no level controls. Although the DCX2496 crossover has gain and
attenuation available in the digital domain on each
channel, I added analog level controls to the DH500.
I also placed two independent potentiometers on
EP
R
IN
T
ax
R
Photo 5: The completed
speaker is shown without
the grille.
Sensitivity Issues
Many speaker designers do not address the
effects of sensitivity on dynamic range because
they cannot determine what amplification will be
used with a particular design. They are also faced
with Hoffman’s Iron Law. Named after J. A. Hoffman,
one of the founders of KLH, the law states that small
bass enclosure size, low reproduced frequency, and
high sensitivities form three branches of a triumvirate. With any design, you can choose any two
branches, but you cannot achieve all three. Small
enclosure size has become an important consideration in modern designs as well as the demand for
T
IN
EP
R
R
Installs Anywhere.
Outperforms Everything.
The groundbreaking DriveCore™ Install (DCi) Amplifier Series from Crown
delivers pure, unsurpassed power for every performance. And powering every
DCi – like the new 4 | 1250 – is our unique DriveCore™ Technology, which replaces up to 500 parts
with one small, extremely efficient chip. Engineered for analog or digital transport applications
and providing higher channel density, DCi produces sonic power that’s simply beyond compare.
Learn more today by visiting crownaudio.com.
archimedia.harman.com
Learn more
about GreenEdge™.
2013 HARMAN
International
Industries,
audioXpress. Reprinted by permission. For subscription information, call 800.269.6301, or visit www.audioxpress.com. Entire®contents
copyright
© Segment
LLC. All Incorporated
rights reserved.
ax
You Can DIY!
Acoustic Power vs. Drive Power
Dayton UM12-22 Woofer
Heil Air Motion
Transformer (AMT)
Power
Watts (W)
Acoustic
Output (dB)
Power
Watts (W)
Acoustic
Output (dB)
Power
Watts (W)
84 dB
1W
89 dB
1W
92 dB
1W
87 dB
2W
92 dB
2W
95 dB
2W
4W
90 dB
4W
95 dB
4W
98 dB
93 dB
8W
98 dB
8W
101 dB
8W
96 dB
16 W
101 dB
16 W
104 dB
16 W
32 W
104 dB
32 W
107 dB
32 W
64 W
107 dB
64 W
110 dB
64 W
105 dB
128 W
110 dB
128 W
113 dB
128 W
108 dB
256 W
113 dB
256 W
111 dB
512 W
114 dB
1,024 W
IN
99 dB
102 dB
T
Acoustic
Output (dB)
Eton 7” Hexacone Midrange
have no idea how much additional power the UM12-22
Ultimax DVC subwoofer would have taken. But I am
almost certain most audiophiles do not have more
than 1,200 W per channel to drive their speakers, which makes that issue a moot point. In addition, the clipping was less of a problem in this triamped system because only the woofer received
the distorted signal. The more critical midrange and
high-frequency drivers were unaffected. If you use
a single amplifier to drive the speaker full range,
the clipping problem would be greatly compounded.
powUntil recently, other than getting a more pow
erful amplifier, there were several ways to prevent
comclipping. You could listen at lower levels, which com
promised the music’s dynamic range. You could use
a compressor, which would provide better audibility
includof low-level signals but squish the music, includ
ing the nonoffending parts and makes it relatively
lifeless. Or, you could use an equalizer to reduce
frethe offending frequencies, which leads to poor fre
quency balance that affects the music at all levels.
Wouldn’t it be great if you could identify the
most common offenders that produced clipping and
reduce the drive level only for those frequencies and
only when they were high enough in amplitude to
syscause a problem? The good news is that such a sys
tem exists—the Behringer DEQ2496 high-precision
digital 24-bit/96-kHz EQ/RTA mastering processor,
which is what I use.
The DEQ2496’s functions enable you to determine
several key signal modifiers and set specific levels
where action is taken to modify the gain applied
to the signal. The control parameters include the
adjusted signal’s modified gain level, the threshold
where the modification begins, the ratio or rate that
the change occurs, the time delay after the trigger
level occurs for action to be taken (Attack), the time
delay after the signal falls back below the threshold
for the modification to cease (Release), and the type
of action, including 6- and 12-dB/octave high- and
low-pass filters or a bandpass filter with adjustable
Qs. All filters have adjustable frequencies.
At first it is hard to realize this function’s power.
Essentially, you can now tailor the drive level of the
music you play to conform to restrictions caused by
your combination of driver sensitivity and available
power. You can also tailor the drive level at different
signal levels to prevent distortion caused by insufficient linear volume displacement in woofers, which
affects virtually every speaker. This is a sophisticated control and requires some understanding to
utilize it to your advantage. In the DCX2496 crossover, there is a single iteration of this function. In
the DEQ2496 equalizer, three individual iterations
are possible at the same time.
Table 1: Each driver generates a sound pressure level (SPL) for a given input drive power.
EP
R
the rear of the chassis. I connected them to the
input jacks and added a dial plate with reference
numbers for repeatability in setting the levels. I
then used pink noise and a sound level meter as a
starting point to set each amplifier’s gain levels. I
adjusted the levels during testing.
Taming the Clipping Problem
R
Previously, I mentioned that on certain well-re
well-recorded cuts I was actually clipping my powerful
Studio Reference amplifier during drum strikes. I
Photo 6: The DEQ function’s first page on limiting amplifier clipping shows the parameters
numerically and graphically.
Photo 7: The DEQ function’s second page shows the attack and release times.
IN
T
Photo 8: The DEQ function’s third page shows the mode employed, the frequency used,
and the bandwidth numerically and graphically.
Photo 9: The DEQ menu screen shows the DEQ’s action as signal levels change with the
music.
signal level goes up, you can watch the gain being
reduced (see Photo 9).
9). You should notice the signal
level approaching –25 dB and the gain decreasing
to around –5 dB.
In the final part of this article series, I will concon
tinue to problem solve issues related not only to
this speaker but most speakers on the market. I
will also address speaker voicing and compensatcompensat
ing curves for various recording shortcomings. ax
R
EP
R
In my case, the problem with the high-level
drum notes was centered at frequencies around
100 Hz. Since they were very transient in nature, it
was not a problem of average power. The problem
was instantaneous demands that drove the amp
into clipping. The goal was to reduce the drive for
a range of frequencies centered on 100 Hz with a
bandwidth wide enough to reduce the problem yet
have minimum impact on other frequencies. I also
wanted to make those changes with a ramp up and
down timeframe to minimize the chance of them
being audible. Since this function works to reduce
gain above a certain drive level, the goal was also
to shape that gain reduction in a way that it was
not abrupt.
To set the parameters, I used an extremely
well-recorded drum set from a sampler provided
by Legacy Audio. It has a wide dynamic range from
light taps on the cymbals to monster drum whacks.
With the volume level set to hear the low-level
sounds, the drum strikes drove the amp into clipping. This is the perfect scenario for DEQ2496 use.
The DEQ2496 function is controlled on three
menu pages. The first enables you to set the amount
of modification you make to the gain and is indicated in tenths of a decibel. I set the value to –9 dB.
The second is the signal level where the correction
begins to take place. I set the value to –30 dB. The
third is a ratio setting that effectively determines
how fast the gain changes from normal to the earlier
modified gain set. The lower the ratio, the slower
the DEQ2496 will make the change. I selected a 1:3
value, which provided a relatively slow rate. The
settings you choose are shown graphically on the
page as they are entered (see Photo 6).
6).
The second page of the menu enables you to set
the attack and release times and change the trigger
point. I set the attack or onset time to 45.75 ms.
I set the release to 247.1 ms. Photo 7 shows the
times set in that menu page.
correcThe third page enables you to choose the correc
tion mode and parameters. In this case, I chose a
bandpass function with a 100-Hz center frequency.
I chose a 0.75-octave bandwidth, which was a good
compromise between high effectiveness and low
audibility. Photo 8 shows the results.
After I chose the settings, I repeated the test at
the same volume level. The DEQ worked as expected,
enabling levels high enough to clearly hear all the
quiet passages with the proper frequency balance
while preventing clipping on the drum strikes. The
difference was impressive. If you watch the first
page of the DEQ2496 menu while the music is playing, you can see vertical bar graphs of the signal
level and gain modification side by side. As the
About the Author
retired IT director. He is
Boston Audio Society, and
the DC Audio DIY group.
He has authored several
articles in professional
audio journals.
Resources
Eton midrange drivers, www.etongmbh.de/en/products/home-hifi/
midrange-bass-midrange.
Legacy Audio, www.legacyaudio.com.
Legacy Audio, Music Sampler Volume 1, Track 12, “Dynamic Drums.”
Sources
ULTRA-DRIVE PRO DCX2496 digital loudspeaker management system and
DEQ2496 processor
Studio Reference amplifier
Crown Audio, Inc. | www.crownaudio.com
DH500 Power amplifier
Hafler | www.hafler.com
HCA 800-II Amplifier
Parasound Products, Inc. | www.parasound.com
audioxpress.com | December 2013 | 55
www.audioxpress.com
www.audioxpress.com
audio xpress
AUDIOXPRE SS | JANUARY 2014
New York State of Audio
135th Audio Engineering Society Convention
Fresh From The Bench
IGS Audio S-Type 500
Audio Electronics
Audio Oscillator and Stereo VU Meter
By Larry Cicchinelli
Standards Review
The New AES67-2013
Audio Network Interoperability Standard
Hollow-State Electronics
The Development
of Tube Guitar Amplifiers
JANUARY 2014
ax
You Can DIY!
Photo 1: The DEQ function’s first page details the woofer’s limit over extension at low frequencies and high power.
By
Thomas
Perazella
Utilizing DSP-Based Processors
EP
R
(United States)
IN
T
Tips to Resurrect a
This third and final segment of the article series on restoring a Heil air motion
transformer (AMT)-based speaker details additional problem solving methods that use
DSP-based processors. The article also discusses final listening and voicing a speaker.
R
Although the Dayton Audio UM12-22 12” Ultimax
driver has exceptional linear excursion capability,
there is only so much output you can get from
one 12” driver at very low frequencies. I have sev
several CDs with extremely low frequencies that were
recorded at high levels.
In particular, I have a CD from the Boston Audio
Society with a recording of Camille Saint-Saëns’s
Organ Symphony #3, which has very high-level
17-Hz pedal notes at several places in the record
recording. If you attempt to play this at realistic levels
with any ordinary speaker, you will surely drive the
woofer out of its linear excursion range and possibly cause the voice coil to bottom out. Normally,
the solution to this problem would be to reduce
the overall levels. However, that will result in a low
overall volume that is quite unlike listening to the
real performance due to the loss of detail. Note,
this is not a case of amplifier clipping.
The acoustical requirement to increase excursion
with decreasing frequency to maintain the same
output level creates a mechanical problem as the
cone/voice coil assembly moves too far. If you have
ever experienced this, you know drivers make a
distinctive noise when striking the back plate that
is quite unmusical and most disconcerting.
To eliminate the problem, I used the Behringer
DEQ2496 high-precision digital 24-bit/96-kHz EQ/
RTA mastering processor. This DEQ2496’s function
is controlled on three menu pages. The first menu
enables you to set the amount of modification you
make to the gain and is indicated in tenths of a
decibel. The second menu enables you to set the
attack and release times and change the trigger
point. The third menu enables you to choose the
correction mode and parameters.
You can use the DEQ2496 functions to determine
several key signal modifiers and set specific levels
where action is taken to modify the gain applied
to the signal. The control parameters include the
adjusted signal’s modified gain level, the threshold
where the modification begins, the ratio or rate
46 | January 2014 | audioxpress.com
IN
T
Photo 2: The DEQ function’s second page provides input screens showing the attack
and release times.
Photo 3: The DEQ function’s third page numerically and graphically shows the mode
employed and slope with the turnover frequency.
purvoice a speaker. Many people think when they pur
chase a speaker the manufacture has already done
that job. If you build your own speakers and you
chose drivers that are flat, you may think you only
have to adjust drive levels and crossover points.
Unfortunately, neither of these statements is
true. First, I have never seen drivers with totally
flat response. Second, radiation pattern differences
posiwill affect the final balance at the listening posi
differtion. Third, microphone frequency, pattern differ
ences, and their placement during the recording
masterhave major effects on the sound. Fourth, master
ing decisions can vastly affect tonal balance. Fifth,
speaker placements in the room and room effects
also have major impacts on the final sound.
One of the easiest ways to demonstrate how
microphone choices affect the final sound is to play
track 5 of the first Stereophile test CD. J. Gordon
Holt, Stereophile’s founder, sat at a desk and read
one of his articles into 18 different microphones,
one at a time. He did not change anything else in
the recording or subsequent processing between
the sections recorded with different microphones.
You will find it difficult to believe it is the same person speaking. The differences are dramatic. If just
the microphone choice makes such a difference,
imagine what some of the other factors can do to
the resulting sound. You can either ignore the differences or try to compensate for them.
It is important to remember that if the recording
process significantly modifies the original sound,
R
EP
R
that the change occurs, the time delay after the
trigger level occurs for action to be taken (Attack),
the time delay after the signal falls back below the
threshold for the modification to cease (Release),
and the type of action.
Controlling driver excursion is another excellent
use of the DEQ function. In this case, I used a highpass function with a –6-dB gain reduction, a –31-dB
threshold level, and a 1:3 ratio (see Photo 1).
Since this iteration of the DEQ function was set
for a lower frequency then longer attack and release
times could be used. They were 34.86 and 304.2 ms,
respectively (see Photo 2).
On menu page three, I selected the L12 mode for
the high-pass function with a 12-dB/octave slope. At
first, Behringer’s nomenclature is confusing, since we
always think of a high- or low-pass filter as having
unity gain except for the modified areas, which have
their gain lowered. However, in this function, you can
decrease or increase the gain.
The nomenclature makes more sense if you think
of increasing the gain. In this case, increasing the
gain in the lower frequency range would create a
“low-pass” device. However, I was lowering the gain
in the low frequencies, making the “low-pass” function actually more consistent with generally accepted
notions of a high pass.
Still confused? Fortunately, the parameters are
graphically represented so what you select is immediimmediately apparent. I chose a 39.9-Hz frequency. Photo 3
shows the resulting correction.
Testing after I implemented this DEQ function
resulted in a normal listening level for the “Organ
Symphony #3,” with no ugly noises coming from the
woofer during the passages with the heavy pedal
notes. The audible improvements this function makes
are substantial, enabling you to enjoy the music withwithout annoying distractions. Although the real answer
to accurately reproducing these types of recordings
requires several drivers and a lot of amplifier power,
ampliI find the ability to use normal speakers and ampli
fiers to achieve most of what the music contains
astounding. You have to hear the results to believe
it. You do make compromises, but they are a fraction
of what you would have to do without DEQ.
Voicing the Speaker
For the sake of this project, I stretched the traditional definition of speaker voicing to encompass
adjustments that affected many segments of the
chain. The common definition of speaker voicing covers crossover points, crossover slopes, and drive levels to individual drivers. Box size, speaker positioning, and baffle shapes are also part of the equation.
I have been asked several times why you have to
audioxpress.com | January 2014 | 47
You Can DIY!
narrow and sharp frequency bands. They are caused
by energy at that frequency, reflecting off multiple surfaces. This results in the combinations of
reflections being out of phase, causing a drop in
levels. As you put more energy into the problem
frequency, the reflected energy from various surfaces gets equally greater and the cancellations still
occur. When you do this, you also throw the balance
off for all the other non-canceling positions in the
listening area. Probably the best way to avoid this
situation is to move the speakers or the listening
position and to judiciously use room treatments.
flexiDuring the speaker voicing process use a flexi
ble, low-distortion, wide-dynamic range equalizer.
equalPrior to digital signal processing high-quality equal
izers were expensive. Now, it is possible to get a
digital equalizer that offers more flexibility than
ever at prices that are within almost every budget
range. In addition to equalization, some of them
also include signal generators, real-time analyzers,
compressors, limiters, dynamic EQ, and more. For
years, I have been using Behringer equalizers with
great success in all my systems. With this project,
I used the Behringer DEQ2496 mentioned earlier.
I used Liberty Instruments’s application Liberty
Audiosuite (LAUD) for PCs during the first phase
in which quasi-anechoic measurements are made.
An ACO Pacific 7012 microphone capsule feeding a
model 4012 preamplifier and a PS9200 power supply/interface provided the input. To protect the PC,
I fed output from the microphone interface through
a custom interface I built and described in Speaker
Builder (see Resources).
The measurement configuration consisted of one
channel with the microphone located 1 m from the
speaker on axis at a 44” height (i.e., approximately
ear level when seated). Using a tri-amped setup with
an electronic crossover enables you to easily switch
from measuring the overall speaker to separately
measuring each driver by using the mute buttons
on the crossover for the appropriate channels.
Using the Behringer DSC2496 Ultra Drive crossover,
I selected a 24-dB/octave slope and revised 176- and
1,500-Hz crossover points. Next, I took a measurement
of the complete speaker with no EQ. Note that these
measurements are raw with no smoothing. Figure 1
shows the results. The responses when using the
24- and 48-dB slopes were similar but a 12-dB slope
showed a greater suckout in the mid- to high-interface region and was somewhat rougher.
The steep slopes of a 24- and 48-dB/octave iteration are more effective at eliminating interference
between drivers located at different positions. When
using a 12-dB/octave slope, there is more overlap
between drivers with resulting cancellations and
there is no practical way to entirely retrieve the
original. What you can do is make the resulting
sound more to your liking.
I began the voicing process by trying to get the
speaker as flat as possible on axis with quasi-anechoic
testing. Then, I made adjustments to the balance
at the listening position. In the second phase, some
of the adjustments begin to correct for room and
radiation pattern effects.
A common trap when voicing a speaker is to try
to eliminate all room effects. It is just not possible.
One of the worst problems is cancellations at certain
R
EP
R
Figure 1: The completed
speaker’s quasi-anechoic
response is shown after
the appropriate drive levels
were set with 24-dB/octave
slopes, but before any EQ
was done.
IN
T
ax
Photo 4: The final Parametric EQ in data form was applied to the completed speaker to
achieve the flattest quasi-anechoic response.
Photo 5: This is the final Parametric EQ in graphical form that was applied to the
completed speaker to achieve the flattest quasi-anechoic response.
T
IN
in the recording/mastering process. In theory, you
could produce a separate compensating curve for
every cut on every record but you would quickly go
crazy in the process. I have found that producing
about five to 10 curves will suit the vast majority
of recordings.
musiTo test the speakers, I have a group of musi
cal test recordings that I have used for years.
Recently, I added several new selections to that
list as they became available. To streamline the
process, I burned four CDs with the individual tracks
classiI wanted. The CDs included pop, rock, jazz, classi
cal, and country selections. Some of the music also
contained sounds designed to determine transient
Figure 2: The completed
speaker’s quasi-anechoic
response is shown after
the appropriate drive
levels were set with 24-dB/
octave slopes with the final
parametric EQ.
Figure 3: The quasi-anechoic
response of the Eton driven
full range with no EQ shows
a ragged peak between
2,500 and 5,000 Hz.
R
EP
R
reinforcements at different positions as you move
along the axis between drivers.
Next, I used the DEQ2496 to correct the frequency
response. The digital electronic equalizer provides the
option to use several parametric EQ channels. Being
able to break away from fixed graphic type effects
enables you to make finer adjustments across the
entire frequency range.
For this exercise, I used the DEQ2496’s 10 parametric bands to produce a fairly flat curve. Photo 4 shows
the actual applied corrections in data form. Photo 5
shows the same EQ in graphical form. The curve‘s
complex shape is apparent and would be extremely
difficult if not impossible to achieve using passive
crossover components or a standard graphic equalizer. If you look at each section’s bandwidth values,
you will notice they vary and are fairly broad from
0.5 to 2 octaves. This type of overlap enables you to
generate a complex curve.
Overall, the equalized response at 24 dB/octave is
fairly flat from the lower measurement limit of about
300 Hz to 20 kHz. Figure 2 shows the exception, with
two suckouts around 12 kHz and 15 kHz. They appear
to be cavity resonances due to the Heil AMT’s construction. I was able to slightly reduce the effect, but with
suckouts that result from cancellations, you cannot
totally remove them with more drive.
As with room problems, if you increase the drive at
a suckout frequency, the higher output reflected from
the housing’s offending sections increase proportionproportionately and continue to produce cancellations. The only
solution is to change the physical construction that
caused the interfering reflections or to add absorptive
material. I added damping material inside the AMT
housing but it had a limited effect. It is interesting
to note the current AMT’s housing has a more open
design and some of the cavities have been eliminated,
possibly eliminating this suckout.
Earlier, I mentioned the Eton midrange had some
response irregularities as its frequency increased.
Figure 3 shows the driver’s raw response when
mounted in free air. Between 2,500 and 5,000 Hz,
there is a significant ragged peak. It appears to be
the reflected energy from the basket, which will not
only change the timbre but also cause congestion.
Figure 4 shows the comparison with the same driver
on the baffle with the 24-dB/octave crossover slope
and the speaker’s final EQ applied. The major problem in the 2.5-to-5-kHz region effectively disappears.
You would not be able to achieve that result with a
shallow slope in your crossover.
Music Curves
Finally, I produced a set of curves that would
compensate for some of the worst deviations found
You Can DIY!
part in the listening. Individual musical tastes differ, but sometimes you can ferret out a problem
that all people recognize. In this test, I used the
DCX2496 connected to my laptop to switch between
different crossover slopes—while keeping all other
parameters the same. The listeners determined
that 48 dB/octave was not optimal. This surprised
me because in my main system the 48-dB slope
is clearly superior. It provides the greatest clarity,
stage size, and smoothness. My only guess is that
the controlled and consistent radiation patterns of
that system’s drivers lend themselves to the higher
slopes.
The 24-dB/octave slope worked well. However, the
12 dB/octave had some serious shortcomings. One of
the most noticeable characteristics was a shrinking
of the sound stage as the slopes decreased. At the
12 dB/octave, there was a considerable reduction in
the soundstage’s width and depth as well as a loss
of detail. I decided to stay with the 24-dB/octave
slope. The 176- and 1,500-Hz crossover frequencies
eliminated some of the earlier problems I noticed,
so they remained.
advanOne of the Behringer DEQ2496’s great advan
Equaltages is a function called “Virtual Paragraphic Equal
izer” (VPQ). The function enables you to adjust the
effective Q or bandwidth of each individual frequency in the graphic equalizer from a default onethird octave to 59/3 octave in two-third octave steps,
which represents one-third octave on both sides of
the center frequency.
By adjusting the bandwidth to a range around
3 octaves as a start, you can move the EQ’s center frequency up or down with a simple one knob
adjustment while listening to the music. The function
enables you to rapidly locate the center frequency
of the problem compared to the stark effects created by gross high Q adjustments, which can be
misleading. For example, if you have a sound that
is too forward in the presence region, you can start
with a 3-dB cut at a center frequency around 3 kHz
and use a bandwidth of 3 (see Photo 6).
By adjusting one knob, you can move it back and
forth in one-third octave increments until you find
the spot that sounds best (e.g., 2 kHz). I did this several times with different pieces and created curves
that seemed like reasonable compromises for each
of them. Naming them for the problems they corrected (e.g., Pop Bright, Presence, Honky, etc.) made
them easy to remember. Once you save them, you
can simply turn one knob on the equalizer to recall
all your adjustments. Then, you are ready to have a
more pleasurable listening experience.
When you use an equalizer to make compensating curves, resist the temptation to use a lot
IN
T
ax
Figure 4: The Eton’s quasi-anechoic response with 24-dB/octave shows crossover slopes at
176 and 1,500 Hz with compensating EQ.
EP
R
response, dynamic range, frequency range, sound
stage and depth, and definition.
When implementing listening tests, you can
improve your results and make the process easeasier by finding the correct playback level for each
track
ack before it is seriously auditioned. If the level
is too low, subtle details will be lost. If it is too
loud, it will sound unnatural regardless of set
settings. I ran through each track with the base EQ,
determ
dete
rmined
determined
ined what I felt were the correct levels,
and recorded them in a reference spreadsheet. I
later used that sheet to record the final correc
correction curve and relevant information on the piece.
IIne
Inevitably
nevitably
vitably when conducting listening tests with
others, I always get questions about the music
selections. The spreadsheet is an easy way to
answer the questions and eliminate time looking
for background information.
A few of the local audio club’s members took
About the Author
R
retired IT director. He is
Boston Audio Society, and
the DC Audio DIY group.
He has authored several
articles in professional
audio journals.
Photo 6: The Virtual Paragraphic Equalizer function permits you to set almost any
bandwidth for an equalizer correction. It also enables you to move up and down the
frequency scale using a single knob, while retaining that same bandwidth and levels with
the test music playing.
T. Perazella, “Interface for LAUD Measurement System,” Speaker Builder, 1999.
C. Saint Saëns, “Organ Symphony 3,” Boston Audio Society, Test CD-1, Track 3.
Stereophile, “Why Experts Disagree,” Test CD 1, Track 5, http://ssl.blueearth.net/
primedia/product.php?productid=2&cat=0&bestseller=Y.
Sources
7012 microphone capsule, 4012 preamplifier, and PS9200 power supply/
interface
ACO Pacific, Inc. | www.acopacific.com
ULTRA-DRIVE PRO DCX2496 digital loudspeaker management system and
DEQ2496 processor
Liberty Audiosuite (LAUD
((LAUD)
LAUD)) audio measurement and analysis system for PCs
Liberty Instruments, Inc. | www.libinst.com
EP
R
In a nutshell, was the project successful? I
believe it was. The sound is dynamic, wide and
flat in frequency response, low in distortion, and
provides sound stage width and depth. Transients
are exceptionally good, adding to the sense of realism. Even with the basic design’s limitation—using
one driver for each of three frequency ranges—the
addition of the new woofer plus the application of
corrections provided by the digital equalizer and
crossover improved the sound quality when compared to the original version.
Will this replace my primary system with huge
line arrays that have the same radiation pattern
and large amounts of linear volume displacement?
Resources
T
Final Impressions
No way! But, if I had size and cost restrictions, I
could easily live with these resurrected Heil-based
speakers. Long live DSP! ax
IN
of narrow corrections or corrections that exceed
more than a few decibels. The goal is to eliminate
gross problems while maintaining a realistic sound.
The smoother you make the corrections the better
the resulting sound. Much of the music I listen to
sounds good with the basic setting, but it is helpful
to correct some of the less than great recordings
of music that moves me.
R
Analog Class-D Amplifiers
Anaview’s Class-D amplifiers are available to
discriminating OEMs looking for premium sound
and audio performance.
• On-board Power Supply
• Robust Design with
Advanced Protection
• Compact & Efficient
•
•
•
•
Stby & Aux Voltages
Easy to Integrate
Award Winning Sound
UL Certified
Activate your loudspeakers with Anaview!
SWEDEN | USA | HONG KONG | WWW.ANAVIEW.COM
Anaview is a member of the ETAL Group, AB
off an audioxpress membership
acoustics
audioXpress has been serving up the best in DIY audio for more
than a decade! With an increased focus on professional
audio, acoustics, and audio electronics, audioXpress is
expanding its coverage and content to better serve audiophiles
worldwide. Become a member and gain instant access to
design tips, product reviews, and industry insight.
J O I N T O D AY !
audioxpress.com/reprint25
audio

You Can DIY! - Circuit Cellar

Transcription

Similar documents

Evalyn Frondozo - Rex Book Store

Author, Minister, Speaker, Christian Life and Relationship Coach

Bticino 334202 (5 Wire)

SB-GM-C7CP1/10TW3

Step # 13 Open up TMPGEnc 4.0 Express and click on “Start a new

The Dayton Ultimax UM12-22 woofer has several

PDF

driver protection screen

Cleveland Institute of Music

Summer-Newsletter2014