INNOVATIONS IN AUDIO • AUDIO ELECTRONICS • THE BEST IN DIY AUDIO

www.audioxpress.com

Audio Electronics
AUDIOXPRE SS | JUNE 2015

A Tunable Near-Linear
Phase Analog Crossover
By Vincent Thiernesse

Practical T&M
Sound Cards
for Data Acquisition
in Audio Measurements
By Stuart Yaniger
R&D Stories
Genelec 8351
Acoustically Coaxial SAM System
Rethinking a Three-Way Monitor

Headsets
Electrostatic Headphones
By Mike Klasco and Steve Tatarunis

You Can DIY! Sound Control

PIC-Based DDS Device Gold Line TEF
By Larry Cicchinelli Measurement System
By Richard Honeycutt
You Can DIY!
Build a Single-Ended Questions & Answers
Guitar Tube Amplifier Patrick Donohue,
JUNE 2015
By Costas Sarris HearNotes CEO and Founder
 ! pcX Has Moved !
PLEASE NOTE
Our New Address
The Authority on Hi-Fi DIY

Your #1 Source
for
NEW & NOS Vacuum Tubes
DIY Parts & Components
and
Audiophile Accessories
AZUMA

Sales: www.partsconnexion.com Inquires:

order@partsconnexion.com info@partsconnexion.com
Toll-Free - US/Canada
905-631-5777 1-866-681-9602 905-681-9602
5403 Harvester Rd, Unit 1, Burlington, Ontario, Canada L7L 5J7
Debit, Visa, Mastercard, Amex, PayPal, EMT, Money Order & Bank Draft
ax
June 2015 ISSN 1548-0628
The Best Sound You Will Ever Wear
Recently I wrote an editorial for The Audio Voice (our weekly
www.audioxpress.com
e-newsletter) on “why you need to look at Apple Watch and wear it.”
In that article, I introduced a few of the reasons why we cannot ignore
audioxpress (US ISSN 1548-0628) is published monthly, Apple’s new Watch.
at $50 per year for the US, at $65 per year for Canada, Like the iPhone, it’s not about the object, the new platform, or the
and at $75 per year Foreign/ROW, by Segment LLC, Hugo design. It is all those things put together and the way they interact to
Van haecke, publisher, at 111 Founders Plaza, Suite 904, revolutionize so many different areas, from security to payments, from communications to health
East Hartford, CT 06108, US Periodical Postage paid at monitoring. And that’s just the visible intention—as if it wasn’t enough—on Apple’s road map.
East Hartford, CT, and additional offices. I believe the Apple Watch is just the beginning of a “wearable” revolution in electronics and that
the implications will be massive. Remote control is the most obvious application, available from day
one on the Apple Watch. We already heard from audio companies, such as Apogee Digital, who are
releasing an Apple Watch remote control app for recording with the new Apogee/Sennheiser ClipMic
Head Office:
Digital solution. But we should also think about less obvious things.
Segment LLC
The Apple Watch, in itself, features wireless communications and audio capabilities;
111 Founders Plaza, Suite 904
complementing the computing power we carry in laptops (yes, I think they will still be around),
East Hartford, CT 06108, US
tablets, and smartphones. It can be used as a direct replacement for a mouse, for instance,
Phone: 860-289-0800
allowing simple gesture recognition. We’ve also seen it used for gesture interaction with musical
Fax: 860-461-0450
instruments—check out the Artiphon Instrument, recently funded on Kickstarter (www.artiphon.com).
You will see how it can be used to play “air-violin.”
As biometric devices and sensor-hubs, the implications for wearables are huge, and Apple is
Subscription Management:
right when it decided to start promoting health and fitness applications. And of course, it will be
audioxpress
P.O. Box 462256
huge for authentication, wireless payments, access control, security, and entertainment.
Escondido, CA 92046, US All the companies who have recently developed tablets and smartphones apps to remotely control
Phone: 800-269-6301 their music players, portable wireless speakers, or music servers will have new options because the
E-mail: audioxpress@pcspublink.com best remote control is something we can wear, not carry in a pocket or place on a tabletop.
Internet: www.audioxpress.com But let’s move on to real audio applications, beginning with audiology. Two major hearing-aid
companies, ReSound and Beltone, recently announced new apps for Apple Watch allowing preset
and personalized listening scenarios, monitoring, control, and calling. Also, Israeli company Glide,
Postmaster: send address changes to: announced an app to deliver live-video messaging on the Apple Watch. The “hands-off” function
audioxpress works with the iPhone, but if it can stream video, it certainly will do better with audio.
P.O. Box 462256
In today’s electronics industry, technology innovations are created in parallel with simple
Escondido, CA 92046, US
evolutionary development. And that’s based mostly on “development platforms.” Examples include
Audinate’s work with the Dante technology and development platform, over which professional
US Advertising: audio companies are creating a completely new range of audio-over-IP (AoIP) applications. Another
Strategic Media Marketing, Inc. example is Valens Semiconductor and its HDBaseT technology, currently powering connectivity in
2 Main Street
residential and commercial AV installations, and replacing multiple cables and interfaces (HDMI,
Gloucester, MA 01930, US
Phone: 978-281-7708 Ethernet, USB, control signals, and up to 100 W of power) with a single Ethernet cable (using just the
Fax: 978-281-7706 physical layer).
E-mail: audioxpress@smmarketing.us Apple Watch is “another platform” extending the company’s ecosystem of operating systems,
Advertising rates and terms available on request. software development tools, APIs, and hardware hosts. Will Apple be the only company leveraging
this strategy? Certainly not, but its lead, which can be seen in the 64-bit architecture and the fact
that the company controls all its hardware, software, and services, will enable Apple to maintain a
significant edge over the competition.
Editorial Inquiries:
For most audio companies, it becomes obvious that there’s plenty to be gained on leveraging that
Send editorial correspondence and
platform, and that’s why the Apple Watch is relevant.
manuscripts to:
audioxpress, Editorial Department
111 Founders Plaza, Suite 904 João Martins
East Hartford, CT 06108, US Editor-in-Chief
E-mail: editor@audioxpress.com

The Team
Legal Notice: Publisher: Hugo Van haecke Editorial Director: Steve Ciarcia
Each design published in audioxpress is the intellectual Editor-in-Chief: João Martins Art Director: KC Prescott
property of its author and is offered to readers for their
personal use only.
Managing Editor: C. J. Abate Customer Service: Debbie Lavoie
Any commercial use of such ideas or designs without prior Associate Editor: Shannon Becker Advertising Coordinator: Kim Hopkins
written permission is an infringement of the copyright
protection of the work of each author. Technical Editor: Jan Didden
Regular Contributors: Bill Christie, Dennis Colin, Joseph D’Appolito, Vance Dickason, Jan
© Segment LLC 2015
Didden, Gary Galo, Chuck Hansen, Richard Honeycutt, Charlie Hughes,
Printed in the US
Mike Klasco, G. R. Koonce, Ward Maas, Miguel Marques, Nelson Pass,
Bill Reeve, Ron Tipton, Steve Tatarunis, David J. Weinberg

4 | June 2015 | audioxpress.com

 OUR NETWORK

SUPPORTING COMPANIES
ACO Pacific, Inc. 59 NTI Americas, Inc. 27

All Electronics Corp. 35 OPPO Digital, Inc. 13

Avel Lindberg, Inc. 43 Parts Connexion 2

Front Panel Express 63 Parts Express 68

Hammond Manufacturing Co. 3 Powersoft 19

Jensen Transformers, Inc. 5 Primacoustic 5

KAB Electro Acoustics 43 SB Acoustics 39

Linear Integrated Systems 58 Solen Electronique, Inc. 57

Menlo Scientific, Ltd. 67 Tymphany 49

Midwest Audiofest 2015 65 Wavecor 55

NOT A SUPPORTING COMPANY YET?

Contact Peter Wostrel (audioxpress@smmarketing.us, Phone 978-281-7708,
Fax 978-281-7706) to reserve your own space for the next edition of our magazine.

COLUMNISTS
Vance Dickason has been working as a professional in the loudspeaker industry since 1974.
He is the author of Loudspeaker Design Cookbook—which is now in its seventh edition and
published in English, French, German, Dutch, Italian, Spanish, and Portuguese—and The
Loudspeaker Recipes. Vance is the editor of Voice Coil: The Periodical for the Loudspeaker
Industry, a monthly publication. Although he has been involved with publishing throughout
his career, he still works as an engineering consultant for a number of loudspeaker
manufacturers.

Dr. Richard Honeycutt fell in love with acoustics when his father brought home a copy of
Leo Beranek’s landmark text on the subject while Richard was in the ninth grade. Richard is
a member of the North Carolina chapter of the Acoustical Society of America. Richard has
his own business involving musical instruments and sound systems. He has been an active
acoustics consultant since he received his PhD in electroacoustics from the Union Institute in
2004. Richard’s work includes architectural acoustics, sound system design, and community
noise analysis.

Mike Klasco is the president of Menlo Scientific, a consulting firm for the loudspeaker
industry, located in Richmond, CA. He is the organizer of the Loudspeaker University
seminars for speaker engineers. Mike specializes in materials and fabrication techniques to
enhance speaker performance.

Steve Tatarunis has been active in the loudspeaker industry since the late 1970s. His
areas of interest include product development and test engineering. He is currently a
support engineer at Listen, in Boston, MA, where he provides front-line technical support
to the SoundCheck test system’s global user base.

audioxpress.com | June 2015 | 5

 Contents

Features
16 Classical Binaural 28 The Right Filter
Experiences By Vincent Thiernesse
By Ron Tipton Build a “tunable near-linear phase
The seventh article in our multi- analog crossover” design without
part series searching for realistic need for precise, expensive
recorded sound focuses on binaural capacitors.
sound.
46 Construct a PIC-Based
20 Genelec 8351 Acoustically Audio DDS Device
Coaxial SAM System By Larry Cicchinelli
Rethinking a Three-Way Monitor Use a PIC microcontroller to build a
direct digital synthesis (DDS) device
By Aki Mäkivirta, Jussi Väisänen, and
with several useful features.
Ilpo Martikainen
Discover this tri-amplified studio
near-field monitor that introduces 54 Build a Single-Ended Guitar
enhanced directivity control at lower Tube Amplifier
frequencies in a relatively small By Costas Sarris
speaker. Turn simple circuits into a small
but efficient single-ended head
amplifier.

6 | June 2015 | audioxpress.com

 Volume 46 – No. 6 June 2015

Columns
Departments
HEADSETS QUESTIONS & ANSWERS
8 Electrostatic Headphones 42 Epiphany Leads to
By Mike Klasco and Steve Tatarunis Innovative Audio Startup 4 From the Editor’s Desk
An Interview with Patrick
5 Client Index
PRACTICAL T&M Donohue, CEO and Founder of
12 Sound Cards for Data HearNotes
66 Industry Calendar
Acquisition in Audio By Shannon Becker
Measurements
By Stuart Yaniger HOLLOW-STATE ELECTRONICS
60 Finding Information on
SOUND CONTROL Vacuum Tubes
36 The Gold Line TEF By Richard Honeycutt
Measurement System
By Richard Honeycutt Websites
audioxpress.com
voicecoilmagazine.com
cc-webshop.com

@audioxp_editor audioxpresscommunity

audioxpress.com | June 2015 | 7

 ax Headsets
Electrostatic
Headphones

In this month’s article, we review the current

state of electrostatic headphone technology
and provide a bit of electrostatic history.

By
Mike Klasco and Steve Tatarunis
(United States)

Conventional headphone drivers,

ribbon planars, and most air-motion
transducers (AMTs) are electrodynamic.
As with most loudspeaker drivers, they
all use a magnetic structure and some
sort of voice coil (the ribbon planar uses
a flat printed circuit coil). On the other
hand, electrostatic headphone drivers
are similar to condenser microphones,
Photo 1: Arthur Janszen, a young naval engineer designed
and work more like a capacitor. Many, an electrostatic tweeter element and made it ready for
but not all, are used in an open-back commercial production. This is an ad for the JansZen
configuration, and most require a high- Electrostatic Tweeter circa 1952. (Photo courtesy of Retro
voltage power supply, although we will Vintage Modern Hi-Fi, itshifi.com)
also explore low power or self-biased
electrets.
With the electrodynamic speaker
d o min atin g t h e m a r ke t, w hy b ot h e r wit h Due to these positive attributes, electrostatics
electrostatics? Realistically, electrostatics require have the inherent ability to produce an extended top-
special amplifiers with some serious bias voltage, end and flat frequency response with low distortion.
which makes mobile operation a challenge, not The benefits of high-resolution sound reproduction
to mention expensive. The answer as to why are audible, measurable, and rarely equaled by few
electrostatics have a place in the sun is the competing technologies (e.g., ribbon planars). For
electrostatics’ combination of desirable properties. readers who want more background information
Electrostatics provide more transparent sound due about electrostatics, audioXpress published a three-
to the diaphragm being uniformly driven over its part series in the July, August, and September 2014
entire surface, low moving mass resulting in “fast” issues (see Resources). For those who want more
response, and very low distortion. extensive information, Segment Publishing offers

8 | June 2015| audioxpress.com

two books—Electrostatic Loudspeaker: Design and
Construction by Ronald Wagner and The Electrostatic
Loudspeaker Design Cookbook by Roger R. Sanders.

Some Electrostatic History

Before World War II, Chester W. Rice and
Edward W. Kellogg at General Electric researched
electrostatic speakers, but the large diaphragm
size needed for low-end response and lack of
appropriate diaphragm materials at the time made
the electrodynamic cone speaker the low-hanging
fruit and electrostatic development languished until
after the war.
In 1947, Arthur Janszen, a young naval engineer,
took part in a research project. The Navy was
interested in developing a better instrument for an external “energizer box” for connection to a Photo 2: Jecklin Float is
testing microphone arrays. Janszen believed that receiver as well as improved comfort, ergonomics, an iteration of the Jecklin
earspeaker frame, designed
electrostats were inherently more linear than cone and performance (see Photo 4).
by Joerg Jecklin.
speakers, so he built a model using a thin plastic Koss continued making the ESP headphone line
diaphragm treated with a conductive coating, which through 1970s, 1980s, and 1990s with one or two
exhibited remarkable phase and amplitude linearity. new electrostatic models produced in each decade.
By 1952, he had an electrostatic tweeter element The current Koss electrostatic model is ESP-950.
ready for commercial production (see Photo 1). It’s the company’s first electrostatic model to be
Although far from the first electrostatic headphone, released with a fully functional amplifier instead of
Janszen’s basic design was a model for what must an energizer and is positioned as a transportable
be the world’s most geeky looking headphone, the system that consists of headphones, an amplifier,
Jecklin Float which was introduced around 1975 various cables, a battery, and wall wart adapters.
(see Photo 2).
Electrostatic headphone history begins with
STAX, Ltd., the most well-known and respected
electrostatic brand (see Photo 3). STAX introduced
the SR-1 headphone back in 1959 and has focused
on electrostatic headphones ever since. Outside of
Japan, the STAX brand was quite obscure in the
1960s. Even in Japan, its customers were limited
to the “audio maniac” contingent.
Koss was one of the main headphone brands in
the 1960s, coining the term “Stereophones,” and
the company really shook up the industry when
it launched the ESP-6 in 1968. This was the first
mass production electrostatic headphone. It sold for
approximately $160 (about $1,100 in today’s dollars),
which is an unheard of price point at the time.
To maintain compatibility with AM/FM receivers
of the day, Koss utilized step-up transformers inside
the earcups. For best results, the headphones
could be used with an adapter (“energizer”) that
enabled them to be hooked up to an amplifier’s
output terminals or to the receiver’s speaker-outputs
(usually the receiver’s second set of speaker outputs).
The ESP-6s sounded pretty good, but they were a
closed-back design, heavy, and uncomfortable. With
positive customer feedback about the sound, but poor Photo 3: The SR-009 is STAX’s new flagship electrostatic headphone. Photo courtesy of
feedback on everything else, new models contained STAX, Ltd.)

audioxpress.com | June 2015 | 9

 ax Headsets
panels that hung on either side of your head. From
that same era, the Marantz SE-1S electrostatic
headphones and companion EE-1 “energizer” were
also quite popular and affordable.

Electrets
In the September 2014 issue of audioXpress, we
explored a class of electrostatics, the self-biased
electret. Electret headphones are a specific class
of electrostatic headphones. Electrets are unique
in the method by which a fixed electrostatic charge
is placed on the diaphragm. It is essentially a static
electric charge. There are two main mechanical
differences between an electrostatic headphone and
an electret electrostatic headphone—the diaphragm
material with a conductive coating and a high-
Photo 4: Koss was one of However at $1,000, it has some formidable voltage supply bias.
the main headphone brands competition. There have been a dozen or so electret
in the 1960s, coining the In the early 1970s, the Koss electrostatics only headphone designs over the years. Sennheiser’s
term “Stereophones.” The
had a couple of true electrostatic contenders for Unipolar 2000 was probably the first open-back
company really shook
competition. One was from Stanton/Pickering, electret headphone on the market in the 1970s.
up the industry when it
launched the ESP-6 in
introduced in 1971, these electrostatics were Japanese brands such as Audio-Technica, Panasonic,
1968. (Photo courtesy of lightweight and wild looking—picture “futuristic Pioneer, Technics, and STAX also offered electret
Preservationsound.com) Tupperware on steroids.” One of this article’s headphones in the 1970s.
authors, Mike Klasco had the opportunity to Early electret headphone drivers did not have
spend some quality time with them. (After school, adequate sensitivity to be driven by the headphone
he worked in an audio store that had them on outputs of receivers (which were notoriously padded
demonstration). down by resistors). The headphones required their
While Pickering had some early experience in own external supply amplifiers and would definitely
the mid-1950s with electrostatic top-end tweeters, not work with the meager output of today’s
the Stanton electrostatic headphones missed the smartphones.
mark. These headphones were lightweight, but so Over the years, there have been quite a few pure
was the sound. They had no dynamic range or any electret headphones, but they have all disappeared,
bass and they soon disappeared from the market. in part, because good results and high sensitivity can
Speaking of “space cadet” headphones, the be obtained with the modern $1 to $3 factory cost
aforementioned Jecklin Float also appeared in headphone electrodynamic drivers. The exception
1975. They were essentially a pair of electrostatic has been the STAX electrets. STAX’s main focus is
true electrostatics. But, it has offered several self-
biased electret headphones over the years and still
offer them in today’s product line. STAX also still
produces the electret headphone drivers for these
designs.
STAX is truly a “condenser-centric” company.
Its first product was a condenser microphone in
1950, followed by a unique condenser phonograph
cartridge in 1952 and an electrostatic tweeter
in 1954. By 1960, STAX was shipping its first
electrostatic headphone, which is the world’s
first electrostatic headphone. Throughout the
1960s, more electrostatic products followed,
including more headphones and several full-range
electrostatic speakers, the ESS series. By the late
Photo 5: Each Sennheiser Orpheus HE90 set comes bundled with a dedicated tube 1970s, STAX had developed and had commercialized
amplifier, the HEV 90. (Photo courtesy of Sennheiser) self-biasing electret phonograph cartridges and

10 | June 2015| audioxpress.com

electret headphones. Following was various open-
backed, closed backed and off-the-ear (Earspeaker)
configurations. More recent developments at STAX
include electrostatic earphones and the acquisition
of the firm by Edifier in China.

Other Notable Products

Perhaps the most legendary electrostatic
headphones are the Sennheiser Orpheus HE90,
which cost $16,000 when they were introduced in
1991 and typically go for double that today on the
used equipment market. The story goes that in
the early 1990s, Sennheiser gave its engineers a
mission: make the best headphones, irrespective
of price and outdo those Japanese guys (STAX) on proprietary dynamic membrane, with stiffness, Photo 6: The Dharma has a
all fronts. The Orpheus HE90 was made in very mass, and internal damping optimized for use with unique two-way (woofer and
limited quantities, with a total of 300 produced. Each ENIGMAcoustics’ electrostatic high-frequency driver. tweeter) hybrid electrostatic
set comes with the HEV 90, a purpose-built tube Transition between the electrostatic and dynamic design. (Photo courtesy of
ENIGMAcoustics)
amplifier (see Photo 5). Sadly, such an ambitious drivers is accomplished via a phase coherent first-
start in electrostatics was followed (and ended) by order crossover.
the Sennheiser HE 60 “baby Orpheus” electrostatics,
which cost much less—and partially because of its Wrap-Up
less than awesome amplifier—did not make quite Electrostatic headphone designs have been with
the same impact on the reviewers nor the market us for more than 50 years. They are still relatively
as its bigger brother. expensive and exist outside of the mainstream of
KingSound, a Hong Kong company known for its dynamic driver consumer headphones, but for those
full-range electrostats, showed its new electrostatic of us who have experienced their open and natural
headphones last year at High End Munich. The series sound, there’s little argument that they represent
included the solid-state M-10 electrostatic amplifier or the “best in class.” There is a strong future potential
the tube variant, the M-20, along with the KS-H1 and for new self-biased electret headphones, as there
the KS-H2 electrostatic headphones. The headphones have been recent advances in this technology (as
are light and comfortable but in $1,000 to $2,000 can be seen in the ENIGMAcoustics products). Future
price range, the headphone plus amplifier has some development work may enable full-range electrets
serious competition. with more excursion enabling less expensive and
The Irvine, CA-based company, ENIGMAcoustics, complex implementations.
is known for its patented self-biased electrostatic Next month, we take a deeper look at the challenges
(SBESL) technology. Article author Mike Klasco involved in using acoustic test instrumentation as a
had a chance to check out its flat-panel speaker tool for the objective evaluation of headphone sound
products at High End Munich in 2014. He received quality. ax
a private demonstration of a headphone prototype
in ENIGMAcoustics’ demonstration suite at the 2015
Consumer Electronics Show (CES), and also at its R&D Resources
M. Klasco and S. Tatarunis, “Electrostatic Speakers (Part 1): Technology and
facility in Taiwan. ENIGMAcoustics recently announced Construction,” audioXpress, July 2014.
the launch of two products in the high-end headphone
category, the Dharma D1000 Hybrid Electrostatic ———, “Electrostatic Speakers (Part 2): Strengths and Design Challenges,”
audioXpress, August 2014.
Headphone and the Athena A1 Triode Vacuum Tube
Hybrid Headphone Amplifier. The bias voltage is ———, “Electrostatic Speakers (Part 3): The Practical Electrets,” audioXpress,
derived from the signal itself in a proprietary manner, September 2014.
making the Dharma “an easy to use, convenient, and R. Sanders, The Electrostatic Loudspeaker Design Cookbook, Audio Amateur
with its light weight, a very comfortable headphone,” Publishing (now Segment, LLC), September 1993. Item AA-BKAA020, www.
according to the company. cc-webshop.com.
The Dharma has a unique two-way (woofer R. Wagner, Electrostatic Loudspeaker: Design and Construction, Audio Amateur
and tweeter) hybrid electrostatic design (see Publishing (now Segment, LLC), March 1993.
Photo 6). Bass and mid-bass are handled by a

audioxpress.com | June 2015 | 11

 ax Practical Test & Measurement
Sound Cards for Data Acquisition
in Audio Measurements

In this column, I will look at some of the available ways to create

a low-cost system (emphasis on “low-cost”) for lab-grade audio
electronics measurements and provide some examples. I will skip the
superb but expensive systems to concentrate on solutions within easy
reach of budget-conscience hobbyists. The goal is to demonstrate
how you can generate useful data, similar to the graphs and charts in
the audio review magazines, but without the expense.

By
Stuart Yaniger
(United States)

I will confess to being a measurement geek. I spectrum analyzer I used as a whippersnapper

firmly believe that anything you can hear, you can (state-of-the-art for its time) would sell for
measure. The challenges lay in choosing appropriate more than $200,000 in today’s currency. It also
measurements, properly performing them, and outweighed me, so it was not the most convenient
making reasonable interpretations. instrument to use. At the expense of adding a lot of
tedium and time to the measurement of frequency
A Bit of History spectra, you could use a somewhat less expensive
The basic measurements in audio electronics and much less bulky wave analyzer (e.g., the
are voltage, current, impedance, noise, power, Hewlett-Packard 3581A), but it still carried a rather
and distortion. Many years ago, when I was young prohibitive cost. Typically, we could only view the
and dinosaurs roamed the earth, the basic tools tools of professionals with burning envy.
for these audio measurements were voltmeters, Likewise, speaker measurements were done
sine and square wave signal generators, and with swept sine waves or (if you were particularly
oscilloscopes. If your test bench was particularly cool) warble tones, point-by-point, graphed by
well equipped, you might have a distortion meter (or hand. These manual methods suffered from the
two since the measurement of harmonic distortion need to improvise an anechoic chamber (and it was
and intermodulation distortion usually required rarely satisfactorily done). High noise levels, and
separate instruments) and a test microphone. With again, time and tedium were major impediments.
those simple tools, audio amateurs performed all Commercial breakthroughs such as Crown’s Tecron
the basic measurements needed to construct the Time Delay Spectrometry (TDS) analyzer or the use
relatively unsophisticated circuits of the day. The of multiplex and impulse techniques were a blessing
measurement process was slow, tedious, and often for professionals, but again, cost was a deterrent
uncertain. for most amateurs.
Professionals used a few more advanced tools With the advent and evolution of the personal
(e.g., impedance bridges, spectrum analyzers, computer (PC), a revolution took place. One of the
and wave analyzers), but for the amateur, these first shots fired was Bill Waslo’s innovative IMP
were hideously expensive. For example, the Nicolet audio analyzer—about 20 years ago. The IMP was

12 | June 2015| audioxpress.com

 ax Practical Test & Measurement
community, but they were expensive and required
some sophisticated user skills in programming them
to get usable measurement results.
A lthough the audio amateur and semi-
professional electronic design and construction
market for test equipment is miniscule, we can often
leverage technology from other, larger markets.
The next step in the revolution was the widespread
introduction of sound cards to PCs, which targeted
computer gamers. The sound cards interfaced the
computer with playback transducers (headphones
and loudspeakers) and microphones.
Some clever people recognized that the sound
cards were, in essence, data acquisition systems
themselves. They could take analog data and convert
them to computer-readable files or take computer-
readable files and convert them to analog outputs.
The development and refinement of the appropriate
software and hardware has vaulted amateur
measurement capabilities into arenas formerly the
exclusive domain of the pros, and inadvertently,
pushed professional audio equipment manufacturers
to seek new performance levels, measurement
capability, and user-friendliness.
C o m p u t e r- a i d e d m e a s u r e m e n t s c a n ,
unfortunately, be an efficient way of quickly
Photo 1: It took some a sophisticated speaker testing system that could amassing huge amounts of nonsense, so I’ll also
surgery to fit a PCI sound test loudspeaker frequency response and complex try to warn you away from the more common pitfalls
card into a low-level impedance using impulse response (later upgraded so that your measurements are more likely to be
computer! I had to add a few to MLS), and connected to a PC via the printer port. meaningful.
additional components to use Recognizing that sophisticated signal processing
a normal-sized PCI card with power was now available on inexpensive chips and Hardware Basics
my low-profile computer.
that much of the cost and bulk of professional There are several “flavors” of sound cards out
spectrum analyzers was the user input interface there (Sound cards intended for the music recording
and display, the IMP cleverly took advantage of market are usually called “audio interfaces,” so don’t
the early PCs’ newly available graphics and input let my use of “sound card” limit you).
capabilities to provide a remarkably sophisticated Built-In Sound Cards: Most modern computers
tool for amateur speaker builders. (In this context, come with built-in sound capability, generally
I have used PC in the general sense, not limited to using on-board chipsets. In my experience, the
specific brands or operating systems, though in this performance of these systems has been mediocre,
case, the IMP was only compatible with IBM-type but it’s easy and quick to check.
computers running DOS.) One problem with a built-in solution is that often
When I first saw Waslo’s article, “The IMP the only accessible analog input is a microphone
Audio Analyzer,” I literally ran to the mailbox to level input that might even be mono, which is the
send off a check for the kit. Finally, some of the case for the Lenovo laptop I used to write this
incredible capabilities I had at work for scientific article. You’ll need a line-level input to really get
measurements could be available to me at home for the most out of your built-in sound card. If that’s
my audio hobby! And, I wasn’t disappointed. The available, a few minutes doing a loopback test (i.e.,
unit greatly increased my measurement capabilities. connecting output to input) with something such as
On a commercial level, data acquisition systems the RightMark Audio Analyzer (which we’ll discuss
(attachments to the computer that converted when we get to software options) will determine
analog measurements into data accessible by the if your equipment has sufficiently low distortion
computer’s CPU for calculation and display) started and noise or whether you’ll have to purchase an
to become widely available to the professional aftermarket solution.

14 | June 2015| audioxpress.com

 Note that for loudspeaker measurements, the
0
sound card performance can be quite a bit worse
than for electronics measurements without messing −20
things up. That’s because loudspeakers tend to have
significantly higher distortion than electronics, and −40
IM Distortion
generally, we’re not looking at very low level signals. −60
So if your primary interest is loudspeaker frequency Amplitude
response measurements, the built-in sound card in (dBFS) −80
a laptop is likely good enough. However, it will still
−100
require an external amplifier, a test mike, and most
likely a mike preamp. −120
PCI Sound Cards: A major step up is a high-
performance aftermarket card that plugs into the −140
0 5,000 10,000 15,000 20,000
PCI slot of a desktop computer. It is often asserted
Frequency
that plunging the sound card into the depths of a (Hz)
computer is a sure-fire recipe for noise pickup.
Maybe so, but I’ve used two different PCI sound Figure 1: This intermodulation (IM) distortion spectrum of a phono preamplifier was
cards in five different computers and only had one acquired using the sound card shown in Photo 1. Full Scale = 4 VRMS.
issue with noise among all those combinations. My
experience may not be universal, though, so see if
you can buy on a return basis. USB 2.0, they are overwhelmingly available only with
The main market for the PCI format is music 48-kHz and 96-kHz sample rates (which I will discuss
composition and gaming, so there’s a wide range of in greater detail later). However, we’re starting to
available performance and pricing. 192-kHz sample see some nice hardware with higher sample-rate
rates are widely available. capabilities starting to come down in price.
Professional Tip: Do not buy a low-profile
computer if you’re planning to fit a normal-sized Article Series Overview
PCI sound card into it. Otherwise, your setup may In coming months, I’ll provide an overview
look like mine (see Photo 1). Despite the horrible of some of the hardware and software options.
appearance, this actually worked pretty well. Following that, we will consider what external
A s an example, Figure 1 shows the hardware it takes to translate that into a useful
intermodulation performance of a prototype phono instrument. Along the way, I will provide some
stage with the spectrum acquired via the sound examples of actual measurements and what it took
card. Note the nice quiet noise floor. The new PCI to obtain them. I’ll also discuss ways to design and
Express (PCIe) cards are suitably low profile and build your own simple interfaces and jigs to do
will obviate the need for the sort of radical surgery common measurements.
I had to perform. In addition, I’ll review some specific pieces of
External Sound Cards: Most of these are made commercially available software and hardware so
for musicians and bedroom recording. That means you can get a good idea of the trade-offs and choose
they’re mass produced at a relatively low cost. what’s most appropriate for the measurements you
Putting aside audiophile snobbery, this is not a bug, want to do. Finally, we’ll consider some advanced
it’s a feature. Communication with the computer issues in measurement and how they can impact
is most commonly through the USB port, though your results and the way the data are presented.
there’s a few Firewire and Thunderbolt interfaces Next month: Choosing a sound card and some
out there. computer tips. ax
The potential upside of going external is removing
the sound card’s potential of noise contamination
Resources
from being inside the computer case. There’s also H. Biering, “Measurement of Loudspeaker and Microphone Performance Using
superior portability, since you’re not tied down to Dual-Channel FFT Analysis,” Application note for the Brüel & Kjær 2032, 1984.
a desktop machine as you are with PCI interfaces.
The downside is potential noise and ground-loops B. Waslo, “The IMP Analyzer,” Speaker Builder, January–April 1993.
from USB or Firewire power. Because of the intended
market for external sound cards (amateur recording) ———, “The IMP Goes MLS,” Speaker Builder, June 1993.
and communications speed limitations with USB and

audioxpress.com | June 2015 | 15

 ax It’s About the Sound

Classical Binaural
Experiences
This is the seventh article in our multi-part series
that takes readers on a continuing quest for realistic
recorded sound. The series, which began in the
December 2014 issue with mono, then stereo, and
ambisonic sound, and focuses this month on
binaural sound.

By Photo 1: The Neumann KU 100 Dummy Head

Ron Tipton Microphone is a replica of the human head with a
microphone built into each ear.
(United States)

Classical binaural recordings are made with a of a head model is the Neumann KU 100 Dummy
pair of microphones that simulates a human head. Head Microphone (shown in Photo 1). It is a replica
In some cases a “head” model is used. One example of the human head with a microphone built into each
ear. Another example is the Kall binaural transducer
shown in Photo 2.
A less costly approach is a pair of small
microphones that plug into your ears such as the
Sound Professionals model SP-TFB-2-BEC (see
Photo 3). An even less expensive option is a pair of
in-ear Roland CS-10EM microphones. In all cases the
microphones hear and record what you as a listener
hears. That is, all sound that comes from the front,
sides and rear.
When the recording is played back through a pair
of headphones, the listener hears a reproduction of the
original sound with all of the spatial location information
preserved. As an example, I have included the classic2.
wav file (which can be found in the Supplementary
Material section of the audioXpress website). It is a
classical binaural recording from the Kall Binaural Audio
Photo 2: The Kall binaural transducer simulates human hearing. The head has a special website. The recording demonstrates the 3-D effect
microphone built into each ear canal. (Photo courtesy of Kall Binaural Audio) when listened to using headphones.

16 | June 2015 | audioxpress.com

Head Models
But first, the “head” model needs some
explanation. It refers to the Head-Related Transfer
Function (HRTF), a model that describes how a sound
from a specific point will arrive at the ears. We humans
estimate the location of a sound source by comparing
the sound’s arrival times and loudness as heard by
our two ears. Because our ears are so close together,
it would seem at first glance the sound difference
would be negligible. However, our auditory sense has
evolved to the point where we are adept at 3-D sound
location. Each person has a different pair of HRTFs,
one for each ear, but it’s possible to construct an
average or generic HRTF that is reasonably close for Photo 3: The Sound Professionals SP-TFB-2-BEC binaural microphone places the mic
element next to the ear canal, within the Pinna, for the most realistic recording possible
the majority of the population.
from the perspective of the person wearing them.
An HRTF involves a mathematical relationship
between the input and the output functions. If we
have an input and an HRTF, we can produce an output.
Looking at Photo 4, we see a group of loudspeakers is definitely not simple (see Photo 5). It is priced at
around an HRTF head. Assume the loudspeakers $129.95 from Sweetwater, but a demo is available.
are reproducing a concert where each loudspeaker I’ve included a demo copy along with some useful
position was occupied by a microphone during the information and two example MP3 files in the
performance. By recording the head output, we have panorama.zip file in the Supplementary Material. The
a two-channel copy of the multichannel pickup with reviews of this plug-in are mixed. With this many
all of its spatial information intact. The number of controls to adjust, it’s apparently easy to go wrong
channels (loudspeakers) in the photo is perhaps an and it seems the supplied presets are not good places
extreme case, but the setup works for any number of to start. It is generally agreed that your input stereo
channels, provided the microphone and loudspeaker file should be at least 24-bits at 96 ksps to minimize
positions match. the degradation inherent in the processing. Maybe
you can have too much of a good thing.
Converting to Classical Binaural
Longcat Audio Technologies developed a digital A Binaural Room Simulator
audio workstation (DAW) VST plug-in that takes a Jeroen Breebaart took a different approach to
stereo input file and converts it to classical binaural this conversion when he wrote the Isone Pro and Pro
using a “generic” head model. The plug-in is not
available but the trial version, which works for 30
days, is included in the Supplementary Material in the
longcat.zip file along with a longcat-h3d.mp3, which is
an example conversion from Longcat Audio’s website.
I recommend listening to it through headphones.
Bauer stereophonic to binaural DSP is a free VST
plug-in for creating a binaural file from a stereo file
for headphone listening. It, too, uses a generic head
model. I have tested it in Sony Audio Studio and I
prefer the “C.Moy” setting of the Default slider. It adds
some crossfeed between the right and left channels,
which seems to make the sound more lifelike. This
plug-in, a webpage (htm) file with some interesting
information about the plug-in and a converted MP3
file are included in bs2b.zip in the Supplementary
Material. The htm file describes the plug-in theory
with some nice details of how the control settings
affect the sound. Photo 4: The many loudspeakers grouped around a HRTF head, enable us to convert
The bs2b plug-in is simple, while the Panorama an array of sound sources to a two-channel binaural output. (Photo courtesy of NIRO
plug-in from Wave Arts with its 24 control knobs, Surround Sound Systems, www.niro1.com)

audioxpress.com | June 2015 | 17

 ax It’s About the Sound

rone-apache-binaural. The listening experience can

be excellent with the lower bandwidth needed by the
music-only format.
The British Broadcasting Corp. (BBC) does
occasional headphone binaural programming on its
Radio3 website (www.bbc.co.uk/rd/projects/binaural-
broadcasting). You will find an immense amount of
information about its ongoing binaural projects but

Project Files
Photo 5: The 24 control knobs on the Panorama plug-in are not easy to adjust for a
To download additional material and files, visit
pleasing result. The manual explains each knob group, but the user is left to experiment. http://audioxpress.com/page/audioXpress-
Supplementary-Material.html.
Surround plug-ins. He used a binaural room simulator
(see Photo 6). I’ve had fun with his different presets
using Audio Studio as the VST host. This plug-in is Resources
no longer available but the demo (included in the Bowers & Wilkins, www.bowers-wilkins.com/
Supplementary Material in the isone.zip file) seems Society_of_Sound/Society_of_Sound/Music/sos-
binaural-recordings.html.
to run fine. Updated versions are available from
ToneBoosters as TB Isone and Isone Surround. HDTracks, www.hdtracks.com.
I’ve included the Windows XP demo version in the
Supplementary Material and the Windows 7 (and Mac Kall Binaural Audio, www.kallbinauralaudio.com.
versions) can be downloaded from the ToneBoosters
Last.fm, www.last.fm/tag/binaural.
website. A license for the ToneBoosters versions is
reasonably priced at $25. Sony Audio Studio, www.sony.com.

Binaural Music Sources SoundCloud, soundcloud.com/

groups/3d-audio-binaural.
Radio France is providing binaural music for
headphones through the www.bili-project.org Sweetwater Corp., www.sweetwater.com.
website. I watched (it’s also a video) and listened
to Claudio Monteverdi’s Vespers, the binaural effect ToneBoosters, www.toneboosters.com.
was pleasing through headphones and even enjoyable
through loudspeakers, although it was reduced to just Veclip.com, www.veclip.com/tag/audio-binaural.
stereo. If you have a slower-speed Internet connection,
Wave Arts Inc., www.wavearts.com.
you will get better results going to www.lemouv.fr/

Sources
Koss headphones
Koss Corp. | www.koss.com

Neumann KU 100 Dummy Head Microphone

Neumann | www.neumann.com

CS-10EM Binaural microphones/earphones

Roland Corp. | www.rolandus.com/products/
details/1081

Binaural microphones
Sound Professionals | www.soundprofessionals.
com/binaural_mic.html

bs2b Plug-in
SourceForge | www.sourceforge.net

TDL Model 444A headphone amplifier

TDL Technology, Inc. | www.tdl-tech.com/
data444A.htm
Photo 6: This screenshot of the original Isone Pro plug-in is functionally similar to the
updated ToneBooster’s version with areas for HRTF Adjustment and Room Acoustics setup.

18 | June 2015 | audioxpress.com

not much example music. Clicking on the Radio tab at general, you won’t experience 3-D sound. However,
the top of the home page brings up a recorded music you can come close to the headphone experience About the Author
list but it’s mostly stereo. with a pair of carefully positioned (4” to 8”) monitor Ron Tipton has degrees
in electrical engineering
A web search for Binaural Music Downloads speakers, if you keep your head in the “sweet spot”
from New Mexico State
provides a large collection of sources. I’ve listed a centered between the speakers and 3’ to 4’ in front. University and is retired
few of the more interesting sites in Resources. These Even if the sound is more spatial with headphones, I from an engineering
files are mostly MP3, for a smaller file size, but they are don’t think many people enjoy being tied to them, even position at White
binaural through headphones. A good source for more the wireless variety, so closely spaced loudspeakers Sands Missile Range.
In 1957, he st arted
general binaural information is www.binaural.com. may be more pleasant.
Testronic Development
Processing is available to convert binaural into Laboratory (now TDL
Binaural Listening ordinary stereo while preserving the 3-D sound using Technology) to develop
Because headphones are preferred for listening to normally spaced stereo loudspeakers, but I’m getting audio electronics. He is
classical binaural, I have some comments about what ahead of myself. I’ll be describing the work of Dr. still the TDL president
and principal designer.
works for me. Ear buds are lightweight and convenient Edgar Choueiri and Dr. David Chesky (and others)
but it’s difficult to get good bass response from a and Fraunhofer Cingo in a future audioXpress article
0.25” diameter or smaller diaphragm, regardless of called “Binaural from Loudspeakers.”
what the ads say.
A pair of Koss ESP-6As have a good frequency Author’s Note: In my “Playing with Ambisonics” article
response and come with an individually measured from audioXpress April 2015, my description for
graph of the sound that gets to your ears. However, building a small first-order microphone array failed
they weigh 2 lb and require an amplifier capable of to mention that although the Panasonic WM-61A is
driving their low impedance. (I designed the TDL model an omnidirectional cartridge, mounting them with
444A Headphone Amplifier for low-impedance loads, sealed rears makes them semi-cardioid. Also, their
which includes my ESP-6As.) responses should be matched, but they are sufficiently
You can listen through loudspeakers, but in inexpensive so you can purchase extras. ax

DID
YOU
KNOW
THAT IPALMOD
A 8500W audio amplifier, a correction of the uncertainties
Differential Pressure Sensor, that are typical in any acou-
1 X 8500 W @ 2 Ω
a Zero-Latency DSP and a de- stical system and increasing Advanced technology
dicated transducer: all this in a the “mains input to acoustic for advanced designer
closed-feedback loop. This is output” efficiency.
the IPAL system (Integrated IpalMod, the most effective DIFFERENTIAL
PRESSURE CONTROL

Powered Adaptive Loudspea- systems for the acoustic de-

ker), the revolutionary techno- signer.
logy, introduced by Powersoft,
that allows to arbitrarily modify A
the driver’s Thiele-Small para-
meters, adapting the transdu- SUBWOOFER
cer’s physical characteristics
to the acoustic design. The CAN GO
designer will have full control
over the system reaching un-
BEYOND
paralleled linearity, real-time PHYSICAL? powersoft-audio.com

Powersoft_halfpage_Audioxpress_IpalMod.indd 1 7/14/14 7:28 PM

audioxpress.com | June 2015 | 19

 ax R&D Stories

Genelec 8351 Acoustically

Coaxial SAM System
Rethinking a Three-Way Monitor
Introduced during the 137th Audio Engineering
Society (AES) Convention in Los Angeles, CA, the
new Genelec 8351A Acoustically Coaxial Smart
Active Monitor (SAM) System is a tri-amplified
studio near-field monitor with an ingenious design
that introduces enhanced directivity control at
lower frequencies in a relatively small speaker.
This brilliant design reflects the Finish brand’s
“outside-the-box” thinking and will help improve
reference audio in recording studios.

Introduction by
Photo 1: It is important for a studio monitor to work
João Martins and Vance Dickason equally well in vertical or horizontal orientations. The
Article by 8351 is a significant accomplishment for Genelec.
Aki Mäkivirta, Jussi Väisänen, and
Ilpo Martikainen
(Finland)

The mid- and high-frequency section of this by Genelec’s R&D team and built entirely in-house
three-way design is comprised of a coaxial 5” at the Genelec factory in Iisalmi, Finland, to ensure
midrange with a 0.75” tweeter. The clever part, accuracy and reliability.
however, is the woofer section in which Genelec’s In this article, we share the design process
engineering staff incorporated two 8.5” × 4” slot- as explained by the authors—Aki Mäkivirta,
loaded racetrack-shaped low-frequency transducers Jussi Väisänen, and Ilpo Martikainen—in their
at the top and bottom of the enclosure. While this paper “Design of an Acoustically Coaxial Three-
intuitively looks as if it might not be optimal in both Way Monitor,” which was presented at the 28 th
horizontal and vertical planes, the opposite is true Tonmeistertagung VDT International Convention,
(see Photo 1). November 2014. The Genelec 8351 SAM System
When we look at the horizontal isobaric was initially introduced the prior month, in October
directivity plot shown in Figure 1 and the vertical 2014, at the 137th AES Convention in Los Angeles
isobaric directivity plot shown in Figure 2, they are (see Photo 2).
reasonably matched in performance. Obviously, it
is important for a studio monitor to work equally The Acoustically Coaxial Design
well in both possible orientations. However, The three-way acoustically coaxial monitor uses
most monitors do not, so this is a significant a minimum diffraction co-axial midrange/tweeter
accomplishment for Genelec. driver seated in a directivity control waveguide. The
Every sub-system in the 8351—electronics, the innovative solution allows acoustically concealing
drivers, and the mechanical assembly—is designed the woofers seated in the front enclosure under the

20 | June 2015 | audioxpress.com

waveguide, creating a large continuous front baffle
surface for mid and high frequencies.
The two woofers, placed at the enclosure
ends, acoustically combine to extend controlled
directivity to bass frequencies. This acoustically
concealed woofer system enables a compact
enclosure with controlled directivity across the
full audio band and directivity control on par with
physically larger studio monitors. Horizontal and
vertical characteristics are similar, enabling use in
any orientation. For example, to minimize console
top reflections, the monitor is vertically placed
for high vertical directivity. If the side walls are
close, the monitor can be horizontally mounted,
reducing acoustic interaction with the walls. Other
Figure 1: The 8351 monitor’s horizontal plane directivity (normalized with on-axis) is
room influences can be minimized with smart signal shown.
processing features.
This design responds to the current needs of
today’s recording and mixing environments, which
are shrinking in size and becoming more uneven
in their frequency responses. At the same time,
the production work must be done within the
constraints of tighter budgets, yet with increasing
quality requirements for the end product.
More coloration, larger differences between
r o o m s, m ul tip l e t y p e s of p r o du c tio n an d
repurposing of rooms during the day or week
require reproduction systems with high neutrality
for accurate editing work, and high sound pressure
level (SPL) capability when required (see Photo 3).
Although a two-way active monitor is the
industry’s ubiquitous reliable workhorse, a three- Figure 2: The 8351 monitor’s vertical plane directivity (normalized with on-axis) is
way monitor design presents the performance detailed.
optimum in many ways. A three-way monitor is
needed for applications where:

• Very high-quality audio is required

• Accurate sound stage imaging is wanted
• It is necessary to hear subtle acoustic details,
even at high SPL
• Accurate control of monitor directivity is
necessary

The three-way monitor design enables refined

selection and optimization of individual driver
characteristics (i.e., sensitivity, linearity, output
capacity, and directivity) because of the narrower
frequency range for each driver (see Photo 4).
Optimal crossover frequencies can be chosen based
on the acoustical constraints given by the drivers
and the enclosure characteristics. A larger surface
area of the waveguide can enhance directivity and
achieve the best performance in the presence of
challenging room acoustics typical of small or Photo 2: Genelec’s R&D Director Aki Mákivirta introduces the 8351 monitor at the 137th
acoustically compromised rooms. Audio Engineering Society (AES) Convention in Los Angeles, CA.

audioxpress.com | June 2015 | 21

 ax R&D Stories

Photo 3: The Genelec Benefit of Coaxial Monitor Concept

8351 monitor’s acoustic The coaxial driver arrangement has an important
performance reaches a
acoustic benefit. A multi-way monitor loudspeaker
peak SPL per pair ≥123 dB
has two drivers simultaneously reproducing sound
(referenced at 1 m) and a
maximum SPL (half space, across the crossover range of frequencies. A
on axis, at 1 m, average conventional multi-way design has driver locations
100 Hz to 3 kHz), with a distributed across the front baffle. The summation of
frequency response from the outputs from any two drivers has been designed
32 Hz to 30 kHz (–6 dB), to be in phase in the primary listening direction
38 Hz to 21 kHz (±1.5 dB). (the acoustical axis). When the two outputs are in
phase, the combined system output remains flat. For
off-axis locations, the two drivers are not in phase
because the distances to the drivers have changed
from the design geometry. This can produce strong
About the coloration in audio. The frequency of this coloration
Authors also changes when the listener moves further off
Ilpo Martikainen is axis. Placing drivers coaxially can eliminate this
founder and chairman coloration for off-axis positions as the geometry
at Genelec. Jussi remains the same for the off-axis positions.
Väisänen is lead The crossover performance has been further
acoustic designer at
Genelec. Aki Mäkivirta
improved by time-aligning the two coaxially
is the R&D Director at constructed drivers at the crossover frequency,
Genelec. producing the benefit of maintaining the time
Genelec is a domain waveforms. This is particularly important
manufacturer of for driver constructions with which the physical
active loudspeaker height of the tweeter and midrange drivers are
systems based in
significantly different.
Iisalmi, Finland. It
designs and produces
products especially Evolution of the Coaxial Driver
for professional studio The concept of implementing a multi-way
recording, mixing and driver system coaxially is not new. In the 1940s,
mastering applications,
Altec Lansing launched the 601 coaxial driver with
broadcast, and movie
production. a section horn tweeter coaxially located in a 12”
woofer. This was followed closely by the Tannoy
Dual Concentric design. By the 1970s, the UREI
813 three-way coaxial monitor became popular in
audio monitoring.
When the woofer cone surface forms the
waveguide for the high-frequency radiator,
intermodulation distortion tends to be larger
than it is for designs with a fixed horn located
in the woofer’s apex. The intermodulation can
be reduced by directing low frequencies to a
separate woofer driver, using the coaxial driver as
a midrange-tweeter radiator. The intermodulation
distortion problems typical for these designs can
Photo 4: The 8351 monitor’s be significantly reduced with a physically displaced
response can be adjusted woofer in a three-way arrangement.
using six parametric The fixed horn tweeter coaxial designs suffer
notches, two low-frequency
from diffraction of sound around the horn’s
shelving, two high-frequency
mouth. The sound diffracting around the horn’s
shelving, a 80 ms delay, and
a 30-dB level adjustment.
mouth travels to the woofer cone, then reflects
The connections are 1 x XLR backward and sums with the direct sound with a
analog in, 1 x XLR digital in, fixed delay and frequency specific level, creating
1 x XLR digital out. sound coloration.

22 | June 2015 | audioxpress.com

 When we look at the small three-way monitoring
products available today, we see four primary design
types. The first design type is the conventional non-
coaxial driver arrangement, in which individual
drivers are arranged in the loudspeaker enclosure’s
front baffle, displaced from each other. When
three drivers are crammed onto the compact
front baffle area, typically there is no surface area
left for directivity control in the midrange and
woofer frequencies. It becomes difficult to arrange
the drivers on the front baffle so that crossover
coloration problems at the two crossover frequencies
can be avoided for off-axis sound.
The second type is the approximately coaxial
driver arrangement, which has an island containing
a conventional tweeter and midrange driver on top
of a woofer. In some approximately coaxial systems, Figure 3: The minimum diffraction coaxial (MDC) design—a 130 mm (5”) coaxial driver
all three drivers have been stacked but the driver with midrange and a 19 mm (0.75”) tweeter eliminates all diffraction sources, using
axes may not be coaxial. Approximately coaxial an acoustically optimized seating of the tweeter and a smooth acoustically continuous
surface from the tweeter outward. This design enables good blending to the directivity
designs allow smaller physical size for the monitor’s
control waveguide.
front baffle but usually have problems with the
drivers’ height differences and the delayed addition This sound energy takes time to travel to the
of the diffracted audio to the direct radiation, midrange cone surface and is then reflected back,
coloring on-axis audio. Directivity matching can and eventually sums with the direct tweeter sound
be problematic and cause coloration issues for the a short time later. This introduces SPL direction
off-axis audio. and frequency-dependent variations, making it
Today, the coaxial operating principle is enjoying challenging to achieve flat frequency response on
a degree of renaissance as coaxial drivers can more than one axis for these designs.
implement physically smaller multi-way monitoring Today’s coaxial three-way monitors try to
systems. Both coaxial driver construction principles alleviate the inherent problems of the coaxial
can be found in three-way monitors. drivers by digital signal processing. Although it can
The two coaxial driver technologies used in improve the flatness of the frequency response,
three-way arrangements include variants of signal processing cannot correct for problems
the concentric tweeter-midrange driver and the with directivity control, colorations in the off-axis
horn-loaded tweeter system (fixed tweeter horn) positions, or the tendency of a coaxial driver to
suspended coaxially with the midrange driver. produce distortion.
However, several characteristic problems remain
with three-way monitor designs that use coaxial
drivers. It is difficult to optimally design the seating
and surface shapes of the coaxial driver. This can
lead to poor control of diffraction and colorations.
Although the concentric arrangement tendency
to cause intermodulation distortions can be
alleviated by using a separate woofer, it is usually
not possible to sufficiently optimize the midrange
cone shape to make the midrange work as a high-
performance waveguide for the tweeter. Seating
of the midrange cone to the driver chassis and
Figure 4: Acoustically
the joint between the tweeter and the midrange
concealed woofers radiate
typically introduce diffracting edges in the design.
low frequency through
The fixed horn tweeter approach for a coaxial acoustically optimized
driver reduces the intermodulation problems but has openings in the ends of the
other challenges. There is usually some frequency front baffle. This enables the
dependent amount of sound diffracted around whole front baffle to be used
the edges of the tweeter horn’s mouth opening. as a waveguide.

audioxpress.com | June 2015 | 23

 ax R&D Stories

coaxial (MDC) driver unit, hidden under the front

surface (see Figure 3). The total woofer diaphragm
area is close to that of a 255 mm (10”) woofer.
The two woofers are located under acoustically
optimized extensions of the front baffle. Woofer
radiation happens through the openings, one for
each woofer. This acoustical concealment is the
cornerstone, enabling a compact coaxial monitor with
good directivity control. Two spaced woofers create
acoustic directivity extending to woofer frequencies.
The directivity produced by the concealed woofer
concept is considerable (see Figure 4).
The distance between the centers of the two
woofers is 268 mm. However, the radiation occurs
out of two acoustically optimized slots. These two
slots are spaced with a distance of their acoustic
centers of 380 mm.
This corresponds to 0.56 wavelengths at the
highest operating frequency, which is 490 Hz at
Figure 5: Genelec’s minimum diffraction coaxial (MDC) driver enables excellent integration
the woofer-to-midrange crossover point. This
with the directivity control waveguide (DCW), avoiding diffractions and generating a flat
frequency response on and off axis.
distance is short so the system of two woofers
has directivity characteristics similar to one large
round woofer driver with a diameter of 457 mm
Acoustically Concealed Woofers (18”) in the vertical plane. On the horizontal plane,
Genelec’s novel approach consists of using a new the directivity is determined by the longer axis
acoustically coaxial minimum diffraction three-way diameter of the oval woofers.
monitor loudspeaker design. This design uses the
entire front baffle as a directivity control waveguide Directivity Control
by hiding two woofer drivers under the front baffle. Matching directivities of the drivers in the
The monitor uses two 215 mm × 100 mm (8.5” acoustically coaxial three-way monitor improves
× 4”) woofers displaced out on the two ends of the system performance.
the enclosure front. The woofers are housed in a The woofer system’s combined acoustic axis
shared bass reflex enclosure having a reflex port created by the two spaced drivers radiating though
opening to the enclosure’s back. The total woofer the slot openings coincides with the acoustic axis of
surface area and the magnetic motor capacity have the minimum diffraction tweeter-midrange coaxial
been maximized. The woofers occupy the remaining driver. This matching creates an acoustically coaxial
front baffle surface outside the minimum diffraction three-way system with the added benefit of a woofer
system with controlled directivity (see Figure 5).
The diffraction-free coaxial driver blending with
the directivity control waveguide of the enclosure
front creates a continuous optimized acoustic
surface for the high frequencies radiated by the
tweeter as well as the midrange. The coaxial driver
design does not have any acoustic discontinuity, such
as a gap or a step change of the acoustic surface
profile, either at the joint between the tweeter,
the midrange inside edge, or at the outer edge of
the midrange. The coaxial driver functions without
acoustic diffraction effects through its operating
frequency range.
The large waveguide enables good control of
directivity across the operating frequency range
of the tweeter-midrange system. The waveguide
Figure 6: The diagram shows how the woofer system directivity extends to low geometry has been optimized to minimize sound
frequencies. coloring diffractions.

24 | June 2015 | audioxpress.com

 Figure 7: Frequency
a) b)
responses of the acoustically
coaxial monitor design on
the horizontal plane, on the
acoustical axis, and at 15°,
30°, 45°, and 60° off axis
(a). Typical extension of
the response to frequencies
above the audio range are
shown (b).

The compact three-way design presents a bandwidth of almost 50 kHz, allowing system
waveguide surface area equal to that found in equalization to extend above the audible range (see
large size three-way main studio monitors. The Figure 7). The high corner frequency is typically
directivity control remains down to 300 Hz in 30–40 kHz (±6 dB). The on-axis frequency response
the vertical direction and down to 700 Hz in the shown in Figure 1 and Figure 2 is flat to ±1.5 dB
horizontal direction (see Figure 6). Considering (38 Hz to 21 kHz).
the enclosure’s compact external dimensions, the Because of the large direc tivit y control
monitor has excellent ability to control directivity, waveguide effectively covering the monitor’s entire
typically only found in monitors with significantly
larger enclosure sizes.

Characteristics
The new acoustically coaxial three-way design
uses two woofers 215 x 100 mm (8.5” x 4”) and a
minimum diffraction coaxial driver that contains a
130 mm (5”) midrange and 19 mm (0.75”) aluminum
dome tweeter. These are powered by two 150-W
Class-D amplifiers (a woofer and a midrange)
and one 120-W Class-AB amplifier (tweeter). The
crossover frequencies are 490 Hz and 2.6 kHz.
The external dimensions are similar to a typical
two-way, with a 452 mm (17”) height, a 287 mm
(11”) width, and a 278 mm (11”) depth.
Analog and digital audio sources can be directly
connected to the monitor. The analog input is
capable of accepting signals up to +24 dBu. The
analog input is converted for signal processing and
reproduction. Monitoring of digital audio in AES/
EBU format supports sampling rate up to 192 kHz.
Digital audio is sample rate converted to ensure
synchronization with all sources. An AES/EBU thru
output provides daisy-chaining to more monitors.
The digital audio AES/EBU and analog audio input
signals are filtered with digital signal processing
at an internal 96-kHz sampling rate. There is a
48-kHz-wide audio path from the inputs to the
Figure 8: Multitone signal-to-distortion for case examples of different technologies: (A)
acoustic output. Input-to-output system latency is concentric, (B) conventional non-coaxial, (C) approximately coaxial, (D) fixed tweeter horn
approximately 5 ms in mid and high frequencies. coaxial, and (E) proposed technology. The classical concentric system is showing a fairly
The low corner frequency of the coaxial monitor strong tendency for intermodulation distortion and a low signal-to-distortion ratio. This may
system is 32 Hz. Signal processing has an electronic be due to the driver’s individual characteristics and may not reflect a universal principle.

audioxpress.com | June 2015 | 25

 ax R&D Stories

the frequency band 100 Hz to 3 kHz exceeds 110

dB (referenced to 1 m distance). The long-term
maximum SPL is limited by the overload protection
to about 105 dB SPL.
The acoustically coaxial three-way monitor also
features flexible signal processing tools (Smart
Active Monitoring). These tools add a layer of
intelligence to the monitors. The monitors connect
to a control network. All aspects of the settings in
a multi loudspeaker system can be adjusted at the
listening position. Automated aligning of monitors
and subwoofers with regard to level, timing,
and equalization of room responses can reduce
room influence on sound and improve accuracy of
monitoring.

Performance
The performance of the proposed acoustically
Photo 5: The Genelec 8351A Acoustically Coaxial Smart Active Monitor (SAM) is used in a coaxial monitor construction was evaluated
studio environment (http://parkstudio.se/info.htm) in Sweden, where Stefan Boman, the against the typical performance of the established
engineer and producer and also part owner requested a trial. (Photo courtesy of Lars-Olof implementations of compact three-way monitors.
Janflod) A multi-tone measurement signal was used. Multi-
tone measurements can provide relatively realistic
front baffle, the frequency response neutrality can measurements of the overall linearity because the
also be maintained for the off-axis directions. This signal is a complex multi-frequency signal containing
implies that the color of sound radiated into the a wide bandwidth, essentially presenting a load
room, and audible as the late energy in the room, similar to a realistic wideband audio signal. The
also maintains a relatively neutral character. test signal contains 30 tones distributed evenly on
In a typical monitoring environment, the a log frequency scale across the audible range,
instantaneous peak output SPL exceeds 123 dB from 21 Hz to 20 kHz.
(referenced at 1 m) per a pair of monitors. For The case examples have included the typical
an individual monitor, the maximum short-term technologies currently applied in compact three-way
SPL measured in half space, averaged within monitor designs. The technologies were represented
by case example products of the concentric type
tweeter-midrange system (A), a conventional
separate driver arrangement (B), approximately
Resources coaxial approach with the tweeter located on an
Altec Lansing, “DUPLEX Loudspeaker System,” http://alteclansingunofficial.nlenet. island suspended over the tweeter and a displaced
net/Duplex.html, September 2014.
woofer (C), a fixed horn-type tweeter coaxial
J. Borwick, Loudspeaker and Headphone Handbook (3rd edition), Focal Press, 2001. with the midrange driver (D), and the proposed
acoustically coaxial minimum diffraction driver
M. Dodd, “A Wide Dispersion Constant Directivity Dual Concentric Driver,” system (E) shown in Figure 8. It is worth noting
Convention Paper for the 92th Audio Engineering Society (AES) Convention, Vienna, that in this case study, technologies A and D were in
Austria, March 1992.
fact two-way systems. This may affect the distortion
E. Dupont and S. Lipshitz, “Modeling the Intermodulation Distortion of a Coaxial figures for these two cases.
Loudspeaker,” Journal of the Audio Engineering Society (JAES), 2009. The multitone distortion to audio signal ratio
was measured for each of the case examples and
Genelec, “8351A Three-Amplified SAM Monitor System,” www.genelec.com/ the distortion within the frequency range 100 Hz
products-menu/products-by-series/sam-series/8351a.
to 10 kHz was considered. It is worth noting that
P. Klipsch, “A Note on Modulation Distortion: Coaxial and Spaced Tweeter-Woofer when the frequency range was limited to 500 Hz to
Loudspeaker Systems,” Journal of the Audio Engineering Society (JAES), 1976. 10 kHz, the results remained relatively the same.
The signal level for Cases B to E was 89 dB SPL,
Tannoy, “Dual Concentric Loudspeaker Drive Units,” www.44bx.com/tannoy, and for Case A, 84 dB SPL.
September 2014.
It may also be noted that the conventional
separate driver three-way system did not show any

26 | June 2015 | audioxpress.com

advantage in terms of the intermodulation distortion distance for the proposed design is smaller than
although theoretically, it should have an advantage 0.5 m (1.5’). Freedom of orientation with the very
over the coaxial systems. short minimum listening distance means that an
Approximately coaxial, fixed horn coaxial and the acoustically coaxial three-way monitor can work in
proposed technology were similar in terms of the near-field and mid-field applications.
complex intermodulation performance but will show The acoustically coaxial three-way monitor is
clear differences in terms of the off-axis coloration slightly more directional along its long axis. This
and on-axis frequency response flatness. can provide benefits (e.g., when the console top
As the case study examples are believed to reflection must be minimized). The monitor can be
display state-of-the-art performance, it could be placed vertically, and the higher vertical directivity
established that the proposed method showed reduces the console top reflection level.
favorable performance in terms of distortion Since all the signal processing is done in the
characteristics. digital domain, it allows for excellent frequency
response control. The signal processing can be
Applications tuned for the installation location and helps in
Acoustically coaxial minimum diffraction design system alignment (see Photo 5). An automated
with concealed woofers implies that the three- software-based procedure is available for the
way monitor can be used equally well in either room alignment and equalization. The flexible room
orientation, vertical or horizontal. The acoustically influence compensation reduces room coloration
coaxial three-way monitor’s neutral acoustic and improves accuracy of monitoring.
character is retained for off-axis directions in either The compact acoustically coaxial three-way
orientation. monitor offers a high-quality monitor option for
Coaxial design also enables the listening distance space- or size-limited applications, where normally
to be unexceptionally short. The minimum listening a two-way monitor would be chosen. ax

audioxpress.com | June 2015 | 27

 ax Audio Electronics

The Right Filter

This project details my “tunable near-linear phase analog

crossover” design. Despite the fact that digital filters are more
commonly used because they allow linear phase performance,
I built my device using analog components because many
Photo 1: Here is a look inside
people still prefer analog systems. Theories exist about how to my tunable near-linear phase
perform linear phase crossovers (e.g., the well-known Lipshitz- analog crossover project.

Vanderkooy crossover). The presented method is easy without

need for precise, expensive capacitors. Moreover, the cross-
frequency will be tunable over more than two octaves.

By
Vincent Thiernesse
(France)

In the first article in this series, I describe my do because they require precise components due to
method and give you the tools to build your own the separate concept of the low-pass filter and an
design. In the second article, I will describe the approximation of pure delay. The subject seems to
hardware for the project and the final touches. have been widely abandoned since the emergence of
The hardware project includes 1.8-to-7-kHz digital filters. Nevertheless, recent studies about phase
crossover-frequency tuning with a remote or linearity have renewed the argument for analog filters.
hardware-fixed crossover-frequency from 1.8 to 7 kHz With my design, I intend to prove that my method
and a 3-kHz fixed crossover-frequency configuration provides robust practice results with respect to
with a fifth-order 20-kHz cut-off filter optimized for component tolerances. The basic idea, which is not
high quality. really new, is to simulate finite impulse response (FIR)
digital filters with analog phasor networks where the
Introduction pure delays provided by ideal temporal sampling are
Overtime, digital filters have progressively replaced replaced by analog phasor cells as approximations of
analog filters, but when the two systems do the same pure delays.
thing, you might prefer the analog one for several This approach differs from many classic
reasons—both subjective and objective. The Lipshitz- approaches in which a near-linear phase low-pass
Vanderkooy crossovers are known to be near-phase filter is synthesized with physical independence with
linear. However, they are expensive and difficult to a pure delay (or a close approximation of it) whose

28 | June 2015 | audioxpress.com

 a)
subtraction from the low-pass results in the high-pass.
However, the problem is obtaining practical results
that are “not too far from theory” because the phase
of both structures has to very nearly match.
Here, the low-pass and the high-pass will
emerge from the same structure to ensure a near-
perfect match of the low-pass and the high-pass in
terms of gain flatness, phase linearity, and phase
synchronization between low-pass and high-pass, even
without taking strict design precautions, because the
frequency/phase response won’t depend on perfect-
matching phase between these two structures. Precise
adjustments are also not required (see Photo 1).

Some Analog Methods

First, we will focus our interest in subtractive
analog filters, phase linear or not. Bessel filters are
b)
known to provide low-pass filters with near-linear
phase. When your low-pass Bessel filter is ready,
you need to perform a pure delay or an analog
approximation of it. The subtraction of the pure delay,
or an analog approximation of it, from the near-linear
low-pass filter is supposed to give you the high-pass.
The problem is the way you need to synthesize a
pure delay and how it will match in practice with the
low-pass in term of phase synchronization, which
is required to keep your frequency response close
to theory.
You might use a digital solution to perform the
pure delay. That ensures phase linearity, but it involves
sampling and quantising the signal through which
the high-pass signal must travel. It is also not low
constraint, even with today’s technology.
You might think about an analog approximation
of pure delay, which leads to exponential complexity c)
with respect to the circuit’s desired linear-phase
bandwidth and involves several active components.
In both cases, and especially in the second case, the
problem will be to get the low-pass and the delay to
have almost exactly the same phase, with better than
a 1° similarity. To reach that goal, you’ll have to use
precise and expensive capacitors.
This is probably the main reason why it is hard to
find this type of design. Typically, we find variants of
a filter such as the one described in an Elektor article,
“Active Phase Linear Cross-Over Network,” written
by Elektor lab/editorial staff (September 1987). This
filter, which is not a Linkwitz-Riley filter, is subtractive
but it doesn’t have linear phase characteristics. The
high-pass is obtained by subtracting the low-pass
filter from a phasor but neither is linear in phase.
The phasor is a unity gain phase copy of the low-pass
phase. Further discussion could show how complex Figure 1a: This schematic details a classic substractive filter (nonlinear phase). Included
it would be to have a phasor with the same phase are two simulations: one uses the exact component values (b) and the other uses 5%
as the low-pass, supposedly phase-linear, until the precision capacitors (c).

audioxpress.com | June 2015 | 29

 ax Audio Electronics

crossover frequency (e.g., 2 kHz) and remaining linear with constant acceleration amplitude. Indeed, the
in phase until 20 kHz. signal recorded on a CD is the speaker acceleration
For my project, I don’t use a fourth-order Bessel equivalence. This is required to link with the acoustic
low-pass filter, but I could. As it is fourth order, the damping of high-frequency drivers, which is both
phase rotation is about 180° at the cut-off frequency. representative of electrical/acoustical transduction
So, the pure delay approximation will have to provide and proportional with the squared frequency as long
about 180° phase rotation at 20 kHz. Such a phasor as the wave length is higher than the high-pass
would be at least a 14 th order system with exotic radius.
component values. If you want the tweeter to be protected from
I did a simulation of the circuit described in excessive excursion below its dedicated band-pass—
the Elektor article (mentioned earlier) to show which is necessary to keep it safe and working
how sensitive a simple subtractive filter is to the well—you’ll have to cut more than +40 dB/decade.
capacitor’s precision (see Figure 1). The possible To place the frequency cutoff at a frequency higher
consequences of a 2% deviation of each capacitor than the tweeter’s resonance frequency constitutes
are apparent. However, the overall phase (i.e., of the the beginning of its acoustic band-pass. The 40 dB/
sum of the low-pass and high-pass) is not as greatly decade will only ensure that it will work at constant
affected as the quality of the high-pass. movement amplitude through the low frequencies.
This is an important issue because the high- This means you’ll need high-precision capacitors
pass selectivity must be better than +40 dB/decade. or extensive experimentation to achieve a satisfying
Over its band-pass, a high-frequency driver works result. And we are not even dealing with the linear

All-pass phase response family curves Corrective association example

Influence of corrector on linear band-width Influence of ripple on undulation on band-width

Table 1: Phase correction and near-linear phase bandwidth shows the influences of different corrective actions.

30 | June 2015 | audioxpress.com

phase crossovers yet, just with a “simple” subtractive You’ll recognize a windowed sampled piece of
crossover. cardinal sine. This IR is shown in Figure 2. The
green Gaussian curve only shows similarity. This
Digital Methods is the IR we’ll try to simulate later, hoping for a
This is probably the reason why digital filters naturally smoother shape. It comprises only the
offer an attractive alternative to what promises to central part of a cardinal sine and the chosen
be a time-consuming effort with a less than average window makes it very close to a Gaussian IR. The
probability of success. reasons of that choice will be briefly discussed.
So the digital solution could be ideal as long as You can get a high level of selectivity using a
you are patient enough to wait for the technology to wider part of cardinal sine but it would be a mistake
provide a high enough performance at a low cost. to think it is limitless because it would require a
While waiting for that to happen, you could try to huge number of calculations. What is not more
enhance the quality of your amplifier’s power supply, appreciated than an abundance of op-amps. And
using large capacitors and/or a big transformer, but it would be another mistake to think the final quality
such audiophile “miracles” seem to have dissuaded would be better because high selectivity involves
many “musicoholics” from throwing more gold coins transient oscillations (Gibbs theorem).
at their music systems. Note that, to be a realistic digital filter, this filter
Moreover, the digital solution is not as simple as would need more coefficients to get the crossover-
it seems, even in 2015, because you need a linear frequency as low as 2 kHz. I have previously
phase anti-aliasing filter as well as a linear phase attempted to realize it with 10 TDA1022 analog
smoothing filter in case the source is something delays, but the 70 dB ( N ) and the 0.5% distortion
S
dB

other than a CD player and involves harmonics ratio were definitely prohibitive. Despite that, I will
greater than 20 kHz. show you how such a filter is a good base for a
Now assume that this is the solution you choose. realistic “pseudo-digital” crossover.
Basically, the theory how to proceed is widely
known. You can synthesize a finite impulse response Pseudo-Digital Method
(FIR) filter whose coefficients are close to a piece of For the past 10 years, I have been contemplating
windowed cardinal-sine for the Low-Pass. Here is a and experimenting with a solution that I call the
generic impulse response (IR) formula generating pseudo-digital method. I conducted the experiment
the frequency response shown in Table 1 where k is like I use to (i.e., like someone living far from a
a parameter you have to tune to achieve acceptable community with limited access to academic
rejection: literature).
I was not surprised to learn that similar literature
   π  existed from around the same time, as I am not
1  z −5 10   6  − nk 
sin n 2

h(z) = × + ∑ ×e × z 
−n the only one who had this idea. Nevertheless, it is
6  2 n =0   π   often used with low-pass networks, instead of all-
  n 6  
     pass, as pure delay approximations because they

a) b)

Figure 2: The near Gaussian low-pass impulse response (IR) is shown using a digital IR (a) and a psudeo-digital IR (b).

audioxpress.com | June 2015 | 31

 ax Audio Electronics

matching between two physically isolated systems.

I received good feedback about these filters,
with the exception of the high number of op-amps
and the phase distortion, although noise and
distortion were quite acceptable despite the number
of op-amps. However, I never discovered if any
audio specialists were curious enough to reproduce
it and listen to it.
For this project, I have tried to be more
reasonable in terms of selectivity to reduce
the number of op-amps and use second-order
approximations of pure delays to improve phase
linearity and better quotient:
selectivity
Nb _ OAs

Transmittance is:
2
ω ω
1− 2 j ×m× −
ω0  ω0 
T ( jω ) = 2
ω ω
1+ 2 j ×m× −  
ω0  ω0 

Instead of:
ω
1− j
Figure 3: The structure of often only address low-pass filtering. ω0
T ( jω ) =
the pseudo-digital filter is My first goal was to design easy-to-do high- ω
1+ j
shown in detail. order audio filters with a potentially tunable ω0
crossover frequency. I tried to replace the pure
delays provided by temporal sampling using unity One op-amp is required in both cases.
gain phasor cells (see Figure 3). Tuning the damping factor leads to 1.5-µs time
I was interested in Padé’s approximations of the delay flatness until approximately 0.6 ×f0. This is not
isochronous transmittance of a pure delay as they enough for this application because the op-amps are
can be used with all-pass electronic structures. too important when it comes to achieving acceptable
Pure delay Padé’s approximations show: selectivity.
Note that sampling N times each half-wave of
m 
( − jωTdelay ) 
n

1+ ∑  a cardinal sine for the IR results in the crossover

− jωTdelay
n =1 

( n ×2 )
n
 frequency being the frequency where phase rotation
e  is:
m 
( jωTdelay ) 
n

1+ ∑ 
n =1 

( n × 2n ) 
 π
N
M y i n i t i a l p r o j e c t s u s e d f i r s t- o r d e r
approximations of pure delays and I published per pure delays approximation. Using about m =
three articles in the French magazine Electronique 0.85 damping factor results in:
Pratique. The first and second articles describe my
0.93 f
methodology. The third article describes a “switched f cross  × f0  0
6 6.45
capacitor’s bass tunable filter” where only the low
pass is really affected by switching artifacts. So, if phase is linear until 0.6 ×f0 (see Table 1),
This methodology is interesting due to its I need:
theoretical clarity and the fact that it has low 20
sensitivity to component tolerance. The result is 2 × 0.6
easy, inexpensive, and close to the theory. This samples of each lobe for a 2-kHz cutoff. Restricting it
robustness, with respect to component tolerance, to the central part, which is twice as wide, I needed
is due the fact that you don’t need perfect phase at least 30 phasor cells. This is definitely too much.

32 | June 2015 | audioxpress.com

 So, I had to design a phase corrector placed Tdelay = Tdelay_phasors + Tdelay_corrector_11 + Tdelay_corrector_12
before the series of identical phasors. To do so, I
took my graphing calculator and tried to adjust the with:
coefficients. After only a few hours, I obtained a    ω  ω  
2

 Tdelay _ phasors (ω ) = 1
satisfying corrector design by following the example × 10 × arctan 2 1+2 × j × 0.85865 × -  
 12000 × ω   ω 0  ω 0   
of Chebychev filters with an extrapolation to phase 
    
2
domain (see Table 1).  1  ω   ω
 Tdelay _ corrector _11 (ω ) = ×  2 × arctan 2 1+2 × j × 0.265 ×  - 
I was close to a theoretical formalization. So,  12000 × ω    1.14884 × ω 0   1.14884 × ω 0   
I placed the phase corrector before the filter for 
 1    ω   ω  
2

it to be effective on both low-pass and high-pass. Tdelay _ corrector _12 (ω ) = ×  2 × arctan 2 1+2 × j × 0.1188 ×   -  
12000 × ω    1.53987 × ω 0   1.53987 × ω 0   
The overall time-delay equation, assuming that: 

2000
f 0  12000  ↔ f cross = 1860 Hz We get Tdelay ≈ 240 µs over the linear phase bandwidth.
1.6

Impulse Response Frequency Response

Digital filter

  π    π 
1 6 sin  n ×   1 6 sin  n  f 2  
n2
3  − 23  n ×π × f   n2
 
20 × log10  0.5 + ∑  × e × cos   20 × log10  0.5 × ∑ 
3  − 23
× e × cos  2n × angle  1 + 2 × 0.85 × i ×
f
−  
LP

 3  π   12000  3 n =1  π   12000 120002   


n =1
n×   n  
  3    3 
 
Pseudo-digital filter with our approximation

Table 2: This details the impulse response and frequency response for both the digital filter and the psuedo-digital filter with the upper active octave set at
4 to 8 kHz.

audioxpress.com | June 2015 | 33

 ax Audio Electronics

Conclusion
About the Author Now that we can use an analog system to
Vincent Thiernesse was born in France in 1973. He is a teacher in applied physics. He simulate the behavior of the digital filter previously
has been an electronics hobbyist since the age of 13. Vincent has written seven articles described, we can see the effect of the phasor cells
for the French magazine Electronique Pratique about measuring, audio filtering, and
audio amplification. Vincent has a website www.muselec.fr, where you can find some
phase distortion beyond 0.6 × f0 (while the phase
of his audio creations and music compositions. corrector extends overall phase linearity to 1.6 × f0)
as a blessing because we get only one aliasing of the
frequency response, and this aliasing is rejected far
beyond 20 kHz (see Table 2 and Table 3). Note that
the response of the digital filter repeats itself every
2 π phasing of pure delays and we have restricted
ourselves to 2 π phasing approximations of pure
delays.
Next month, I will describe the rest of the

Impulse Response Frequency Response

Digital filter

  π    π 
6 sin  n × 1 6 sin  n 
1  n2
6  − 30

 n ×π × f  
n2
6  − 30   f f 2  
20 × log10  0.5 + ∑  × e × cos  20 × log10  0.5 × ∑  × e × cos  2n × angle  1 + 2 × 0.85 × i × −  

LP

π   π 12000 120002   
3 n =1   12000   3  
n 
n =1
 n×    
  6    6 
Pseudo-digital filter with our approximation

Table 3: This shows the impulse response and frequency response for both the digital filter and the psuedo-digital filter with the lower octave set at 2 to
4 kHz.

34 | June 2015 | audioxpress.com

project, focusing on crossover-frequency tuning, Author’s Note: I would like to offer a special thanks
and measurements. I will also detail corrector to Stéphane Cnockaert who contacted me to discuss
2, which is simultaneously performing the phase my first filters and gave me probably the best advice
correction and the 20-kHz anti-phase-distortion when working on my personal designs. He said to
filtering. In the meantime, enjoy the filter. Because use as few op-amps as possible; pseudo-order of the
of the limited article size, I did not provide detailed high-pass is better than two; use a near Gaussian
explanations about the implementation. The parts impulse response for the low-pass; have optimal
list, schematics, and PCBs needed to build the main overall filtering to avoid transient artifacts.
board, the digital rotary encoder, and the phasors Also, special thanks to the late André Bustico,
can be found in the Supplementary Materials section who did a simulation of my first filters as soon as I
on the audioXpress website (www.audioxpress.com). designed the circuit and encouraged me to consider
I will also describe the rest of the project with forward-phase linearity. ax
justifications for my technical choices.
My future efforts include enhancing the output
voltage swing, limited to more than ±500 mV with
a crossover-frequency tuning board, depending on
the crossover-frequency value. I will provide precise Project Files
measurements of noise and distortion next month To download additional material and files, visit http://audioxpress.com/page/
as well. For now, I am just saying that the noise DSP audioXpress-Supplementary-Material.html.
is homogeneously spread over the audio band with
an RMS value less than 40 μV. The total harmonic Resource
distortion (THD) is less than 0.1% but I need more “Active Phase Linear Cross-Over Network,” Elektor, September 1987.
precise measurements to give an exact value.

audioxpress.com | June 2015 | 35

 ax Sound Control
The Gold Line TEF
Measurement System

Learn the history of the time

delay spectrometry (TDS)
measurement concept and the
time energy frequency (TEF)
analyzer in this sound control
article. Then, explore one of its
modern-day descendants, the
Gold Line TEF measurement
system.

By
Richard Honeycutt
(United States)

The Invention of TDS and TEF environments, with high noise immunity. Heyser’s
The first computerized acoustical measurement original TDS system consisted of numerous pieces of
systems had their inception in the 1970s, based on standard test equipment (e.g., oscilloscopes, signal
the work of the late Richard C. Heyser, who was then generators, etc.) plus a specialized box of his design,
employed at the Jet Propulsion Laboratory (JPL). which was dubbed the Time-Energy-Frequency (TEF)
Heyser invented a measurement concept called Analyzer. This original system is now housed in
“Time Delay Spectrometry,” or TDS, which enables the Acoustics Department of Columbia College in
quasi-anechoic measurements to be made in echoic Chicago, IL, where it fills three equipment racks.
In 1979, Gerrald Stanley of Crown International
led a team of engineers and technical people who
developed the first dedicated TDS instrument—
the T EF 10. T his advance d “me a sure me nt
computer” included three Z80 microprocessors,
math coprocessors, two 5.25” disc drives, a video
display, a CP/M operating system, and all the
necessary filters, oscillators, preamplifiers, and
support systems. The system was housed in one
approximately 50-lb self-contained unit.

TEF’s Evolution
In 1990, the TEF-20 was introduced. It combined
all of the display and data storage functions onto a
Photo 1: This current TEF 25 standard PC. The TEF line has since been acquired
kit replaces Richard Heyser’s by Gold Line, and the current model is the TEF
original three racks of 25. Heyser’s original three racks of equipment
equipment. have been replaced by a small USB preamp, a test

36 | June 2015 | audioxpress.com

microphone, a handful of cables and adapters, and
a set of software modules (see Photo 1).
The TEF software is collectively called Sound
Lab, and is available in separate modules. The
software modules that are of the most interest
for acoustical measurement are the Noise Level
Analysis (NLA) and TDS modules. The RTA module
provides real-time analysis functionality for sound-
system measurements.
Rather than depend upon a computer sound
card of unknown quality, the current TEF system
includes the TEF 25 USB I/O device with known
premium-quality performance. The TEF 25 is a true
dual-channel device with balanced TRS and XLR
inputs and XLR outputs, excellent phase matching
between channels, 48-V phantom power and a noise Photo 2: This is the TEF opening screen.
floor of –130 dBV. The front panel includes signal-
present, clip, and 48-V LEDs for each input, and
signal LEDs for each output. The TEF 25 is powered which the necessary software modules are present.
by the USB port. An included wall wart is used to The Parameters menu allows the user to define test
provide power for certain 48-V test microphones. parameters for different types of tests.
USB drivers are saved in the computer with the Photo 3 shows the parameters dialog for time-
installation of the Sound Lab software. The user response tests. The Display menu (Main Menu
installs the drivers, following a step-by-step guide toolbar) provides control of screen configurations.
provided with the TEF system. The Tools menu is populated by options depending
on the software modules that are loaded. The
TEF with Sound Lab Reports window allows the user to add a custom
A TEF system equipped with Sound Lab plus the image to printed report pages. The Help button
TDS, NLA, and RTA modules provides the following accesses an html help file that is loaded to the
measurements: user’s computer during the installation process.
Located to the right of center in the Direct Access
• Time Response Toolbar is a smiley-face indicating that good USB
° Basic energy-time curve
° RT60 broadband, full octave, 1.3 octave
° %ALcons
• Frequency response
° RTA
° FFT using TDS to create an anechoic time
window for measurements
° 3-D waterfall plot
• Noise, STI, and RASTI

TEF’s Features
To make measurements using the TEF system,
the user opens the Sound Lab software to the first
screen (see Photo 2). The three toolbars across the
top are designated as the Main Menu, Direct Access
Toolbar, and Measurement Toolbar. In general, any
test function can be accessed in more than one
way, using these toolbars.
Photo 3: The parameters for
The Test menu (Main Menu toolbar) enables the time response tests are set
creation and the execution of custom tests, using via this window, accessible
the FORTH scripting language. The Measure menu through the Parameters
provides one way to execute any measurement for menu.

audioxpress.com | June 2015 | 37

 ax Sound Control
• Time response
• Frequency response
• 3 D (waterfall)
• STI speech intelligibility
• RT60 in full- or 1/3-octave bands
• NLA noise analysis
• RTA functions,
• Additional tests, according to the software
modules that are loaded.

To find the RT60 in octave bands with TEF, the

user first connects a measurement microphone to
the TEF 25’s channel 1 input, and a power amplifier
and test loudspeaker to an output XLR connector.
Photo 4: Input selection and output wave parameters are chosen via the input settings Then, the user selects Mic by pressing the second
dialog. button to the right of the smiley-face on the Direct
Access Toolbar and presses the loudspeaker button
communication with the TEF box exists. To the right on that same toolbar to bring up the Input Settings
of the smiley-face is a button with a loudspeaker dialog. From that dialog, the Ch 1 On and power
logo, which brings up the input settings dialog box check-boxes are checked, and the Ch 2 On button
(see Photo 4). Via this dialog, the channel 1 or the is left unchecked. The measurement and post-
channel 2 input can be turned on or off, or put in processing automatically proceed under the control
loopback mode. Output level and tone type can be of a predefined script. Photo 5 shows the results.
set, and a piano keyboard can be used to select the The 3-D display at left shows the time response for
frequency, using the standard musical pitch scale. each octave band, as well as the broadband response.
The next four buttons on the Main Menu Once the Summary button in the measurement
Toolbar enable selection of inputs for each channel: toolbar has been pressed, a data box appears, as
microphone, internal, or off. To the right of each shown at the upper right. This data box lists the
selection button is a rectangular indicator that broadband RT60 and the RT60 for each octave band
turns yellow to indicate that phantom power for from 125 Hz to 8 kHz. Also shown is the SNR observed
that channel is on. during the measurement. If the SNR is too low to
The Measurement Toolbar provides one-click permit a valid RT60 measurement, the software will
operation of specific tests. These include: display N/A rather than state an erroneous value.
In the example measurement shown, no RT60 was
found for the 250-Hz and 8-kHz bands, either because
of excessive noise in the room or due to insufficient
test signal output from the test loudspeaker. The
RT60 determined by the T10, T20, T30, or ASTM
C-423 method can be displayed, as can the Early
Decay Time (EDT). The broadband RT60, %ALcons,
and direct-to-reverberant ratio are displayed at the
right-hand end of the Measurement Toolbar.
Measuring STI uses a similar procedure, except
the STI button on the Measurement Toolbar is used
instead of the RT60 button. Photo 6 shows the results
of an STI test. The Transmission Index (TI), early RT
(the earliest data point that is –10 dB with respect to
the maximum level, used for %ALcons calculations),
SNR, and noise are displayed for each octave band.
From this data, the STI and the broadband SNR are
calculated and displayed, and the STI classification
(Poor-Fair-Good-Excellent) is shown.

NLA Measurements
Photo 5: The octave-band RT60 measurement results include much information. The NLA module of Sound Lab provides the

38 | June 2015 | audioxpress.com

 ax Sound Control
Photo 6: TEF displays
the results of a speech
The post-process Edit function gives the user the
transmission index (STI) ability to edit an NLA measurement after it has been
measurement in these made by selecting points where extraneous noise
panels. has occurred and muting those points so that they
do not interfere with the measurement. The muted
data remains but it is skipped when calculating
any statistical information about the measurement.
The Import feature enables you to load a
.wav file into Sound Lab and convert it to an NLA
measurement. The Output feature enables you
to save an audio recording from the microphone
at the time of the measurement, which is synced
to the NLA measurement. The audio streaming
function makes it easier to select a time span of a
finished NLA measurement and play back the audio
associated with that selected time span through
ability to analyze noise over a specified time period. the computer speakers or a headset. The Listen
NLA can be set to run for an interval from one function extends this capability to allow you to
minute to 24 hours. Often this type of measurement hear the audio from the microphone during a live
is used for community noise measurements or to NLA measurement session.
document compliance with Occupational Safety and Photo 7 shows a completed NLA measurement.
Health Administration (OSHA) noise requirements in At the bottom is a trace showing the noise level in
the work place. NLA is also useful when designing an office over a one-hour period, with the actual
and installing sound systems in places where clock time shown on the horizontal axis. The
noise is the primary inhibitor to intelligibility (e.g., intervals during which the HVAC system operated
gymnasia, railway stations, etc.). clearly shows, as does one momentary spike during
The latest NLA release (version 6.8.14 or greater) which an object was dropped onto the floor.
is capable of two-channel measurements. Also, The information at the top provides the
a new ability has been added to merge several instantaneous minimum, maximum, and mean
consecutive NLA files into a summary file that noise levels, as well as the levels that are exceeded
contains all the information of the individual files 10%, 50%, and 90% of the time, respectively.
An enhanced version of NLA is available that The frequency weighting is also displayed. The
permits import of noise files in the .wav format. data analysis can be specified by use of the NLA
parameters dialog (see Photo 8).
Selection of the Integration time controls how
the NLA data is calculated. This can be changed at
any time before, during, or after the measurement
has completed. The following choices are provided:

• Leq is the equivalent continuous sound level. The

steady continuous level yields the same total
sound energy over a given period of time as the
varying measured levels.
• Lden is a 24-hour Leq, except 5 dB is added to all
levels measured between 7:00 PM and 10:00 PM
and 10 dB is added to all levels measured between
10:00 PM and 7:00 AM to account for the need
for more quiet during sleep hours.
• Ldn is a 24-hour Leq, except 10 dB is added to all
levels measured between 10:00 PM and 7:00 AM.
• Dose is a rating of the time-weighted acoustic
exposure level that a person receives over a
period of time, usually 8 hours.
Photo 7: A noise level analysis (NLA) measurement shows the noise level vs. time, plus • Linear/Time displays data in time vs. mPascals
statistical data resulting from analysis of the noise. instead of time vs. decibels.

40 | June 2015 | audioxpress.com

 Photo 9: This screen enables
you to enter microphone
calibration information into
Sound Lab.

• Register is a chart of sound pressure levels (SPL) encountered during a

measurement vs. the percentage time that SPL was seen.
• Histogram is a count of the number of times an SPL level was seen during a
measurement.

Microphone Calibration Parameters

To make accurate measurements using any system, microphone calibration
parameters must be used during the calculations. These are entered in TEF Sound
Lab by means of the Input Calibration dialog box (see Photo 9). An auto-calibration
function is also provided, requiring the use of an external acoustical calibrator to
excite the microphone. TEF04 systems are available as individual components or
Photo 8: Parameters for noise analysis are set using this as kits. The standard TEF test microphone is manufactured in the US by Gold Line.
dialog box. For more information, visit www.gold-line.com/tef/t-tefpre.htm. ax

audioXpress 2014
Digital Archive
With this digital subscription, you have
access to all 12 issues of audioXpress 2014
from any computer or tablet at anytime.
Readers can explore project ideas,
bookmark pages, and make annotations
throughout each issue.

audioXpress archive CDs

CD includes all 12 issues for a given
year of audioXpress in PDF format.

cc-webshop.com
Order yours today

audioxpress.com | June 2015 | 41

 ax Questions & Answers
Epiphany Leads to
Innovative
Audio Startup
An Interview with
Patrick Donohue,
CEO and Founder
of HearNotes Patrick Donohue founded
HearNotes in an effort to create
truly WireFree earbuds.

This is the current HearNotes

company and product logo.

By
Shannon Becker
(United States)

SHANNON BECKER: Tell us about yourself. PATRICK: A little-known fact about my background
as a musician is that I’ve also been writing and
PAT R IC K D O N O H U E : I’m o r i g i n a l l y f r o m recording music for nearly two decades. My interest
the Midwest, and although my professional in audio technology initially piqued around age 14
background has primarily revolved around banking when I produced my first album with a 16-track MIDI
and finance, my interest in audio technology stems recorder. Although archaic looking back, it was quite
from my experience as a musician and a classically leading-edge at the time, and I suppose that innate
trained pianist and composer, as well as from curiosity has just stuck with me ever since.
my experience as a national collegiate athlete.
[Patrick is also a two-time Kentucky state tennis SHANNON: Tell us about your company HearNotes
champion.] (www.hearnotes.com).
I hold a bachelor’s degree from Miami University
in Oxford, OH, and began my career in real estate PATRICK: HearNotes is a San Francisco-based
finance, ultimately making my way to JPMorgan startup on a mission to pioneer the evolution of
Chase, where I worked for about five years. Then, portable audio technology and advance the mobile
in 2012, I was promoted to San Francisco, CA, to run music listening experience. The vision was first
one of the firm’s most prominent banking centers in conceptualized on December 14, 2013 in burst of
the Financial District, leading a team of 14 employees profound intuition, triggered as a result of a great
and overseeing a $50 million balance sheet. deal of research I’d already been conducting. It was
a defining moment of clarity around what seemed
SHANNON: How did you get interested in audio my inevitable path towards driving a quantum leap
technology? in audio technology innovation. I walked away from

42 | June 2015 | audioxpress.com

my post at JPMorgan Chase and founded HearNotes
on January 7, 2014.

SHANNON: How did you learn of Kleer technology

and why did you decide to use it in your
headphones?

PATRICK: The industry research that followed my

original “epiphany” admittedly felt a bit like trying
to find a needle in a haystack, without the guarantee
that the needle even existed in the first place. But
that exhaustive research led to the “discovery” upon
which HearNotes is founded—Kleer technology.
Kleer is quite simply one of the industry’s best-
kept secrets for permanently solving the well-known
limitations of Bluetooth. It is the missing evolutionary
link toward building a truly WireFree earbud solution, The HearNotes product
and it represents a quintessential opportunity to family includes two
advance an overly commoditized industry that is in WireFree earbuds, a
dire need of innovation. PATRICK: HearNotes is a truly wirefree mobile music transmitter, and an
solution, which means there is no wire from ear to inductive wireless charging
pad that doubles as a
SHANNON: Tell us about your WireFree Earbuds device or from ear to ear, as there is with other
carrying case.
concept. What makes the concept so unique? earbuds available today. These other so-called

offering an extensive range of ready-to-go

toroidal transformers
to please the ear, but won’t take you for a ride.

Avel Lindberg Inc.

47 South End Plaza, New Milford, CT 06776
p: 860.355.4711 / f: 860.354.8597
sales@avellindberg.com • www.avellindberg.com

audioxpress.com | June 2015 | 43

 ax Questions & Answers
SHANNON: You left your product’s fate in the
hands of the public by posting it on Fundable
(www.fundable.com). Why did you choose the
crowdfunding route and what kind of response
did anticipate?

PATRICK: Fundable, and the crowdfunding route in

general, continues as one element to our strategy.
Late last year, HearNotes was approached by
Fundable to participate in the Staples Crowd2Shelf
Contest because it was suggested that our product
would not only be a strong contender, but it would
also be considered for a rare opportunity to
partner with Staples. We were named a finalist,
and shortly thereafter officially debuted at the 2015
International Consumer Electronics Show (CES) with
tremendous public reception and named Best of
HearNotes premium WireFree earbuds are powered by Kleer wireless transmission Show.
technology from Microchip. Thankfully, our unveiling was blessed with
tremendously positive industry, retail, and consumer
reception and anticipation, one measure of which
“wireless” earbuds and headphones are based on is the more than 70 editorials worldwide and an
Bluetooth technology and, therefore, have inherent estimated 500 million digital impressions. People are
limitations. Specifically, Bluetooth was never designed talking about HearNotes, and we can’t wait to give
for music listening, so utilizing Bluetooth on our people the opportunity to experience the true joy
mobile devices is akin to surfing the Internet with a and freedom of The WireFree Solution that wireless
dial-up modem. And Bluetooth cannot support stereo music should be.
audio output unless there’s a wire (not wireless) that
connects the two earbuds! SHANNON: Was this your company’s first product?
HearNotes is unique in its ability to deliver quality If not, what other products do you offer?
and convenience—pristine, lossless, uncompressed,
high-fidelity audio via the Kleer technology. It is a PATR ICK: HearNotes: The Universal Edition
final departure from those silly, antiquated wires represents the company’s debut generation.
that had us tangled up for so long. HearNotes is the There are plans in the works to develop an array
really truly wireless audio solution the market has of WireFree product offerings. Stay tuned.
been awaiting.
SHANNON: What’s next for HearNotes?
SHANNON: Were there any challenges involved with
the product design? PATRICK: We launched the “We Hear You” Kickstarter
campaign on May 4, 2015 to support the HearNotes
PATRICK: Yes, and I would say that product design effort to redefine mobile music listening and put
represented the fundamental challenge. We first the customer’s experience right at the center of
needed to recruit engineers that truly understood our mission, exactly where it belongs. We are
the nuances of Kleer technology and how to properly super excited about Kickstarter’s vast worldwide
integrate it. Second, given the commoditized nature community of early adopters who are very eager to
of the industry, it wasn’t like we had a playbook support the type of innovation present in HearNotes,
to reference, and previous design attempts were and be among the first who get in on the ground
minimal to non-existent. floor. We anticipate a continued positive response!
The question was never if we could build HearNotes, Any of your readers who want to be an early
but rather how we would design them to be sleek and adopter for HearNotes, should please check us out on
cool enough to facilitate rapid consumer adoption. Kickstarter. For more information about HearNotes,
Thankfully, we have an amazing engineering team visit www.hearnotes.com, Kickstarter, Gethearnotes
and early feedback on design continues to be very Facebook, Gethearnotes Twitter, and HearNotes on
positive. YouTube. ax

44 | June 2015 | audioxpress.com

 ax You Can DIY!

Construct a PIC-Based Audio

DDS Device
I have previously built two direct
digital synthesis (DDS) systems using
a Rabbit microprocessor programmed
in C. I built a PIC-based frequency
counter quite a few years ago but have
not looked at PICs for about 15 years.
So, this time I wanted to see whether
or not I could build one using a PIC and
still include several useful features.

Photo 1: The main menu

By selections are located on the front of the unit.

Larry Cicchinelli
(United States)

This PC-based audio DDS device contains several expansion, although one of them may be set to
convenient features (see the Features List). Photo 1 indicate the marker frequency is being output.
shows the front of the unit displaying the main
menu selections. The three controls are, from left The Menus
to right: the amplitude, the Quadrature Encoder, Navigating the menus is fairly straightforward.
and the four-position joystick. Only one of the LEDs The joystick is used to move the cursor to the
is currently in use and it indicates that the program desired selection. To select the item, simply press
is running. The remaining two LEDs are for future in on the Quadrature Encoder (QE). Pressing in on
the QE is handled as Enter. The current program
Features List does not make use of the joystick’s Enter switch. I
Ÿ Sine wave only—frequency selectable frequencies—the user enters the found this a little bit difficult to operate without also
in 1-Hz steps from 1 Hz to 100 kHz frequencies and duration
activating one of the direction switches.
with a Direct Digital Synthesis (DDS)
resolution of about 0.1 Hz Ÿ User can store all entries and select The only unusual operation is that if the cursor
the mode in which the system starts is on the first entry of the menu, as it is shown in
Ÿ Four line LCD § Menu
Photo 1, and the joystick is moved to the left, you
§ Single frequency
Ÿ Quadrature encoder for changing the § Sweep will exit the current menu and the “parent” menu
frequency § Frequency list will be displayed. The cursor will be placed on the
menu item that had been previously selected.
Ÿ Four-position joystick for navigating Ÿ The amplitude is adjustable via an
the menu and changing the frequency analog potentiometer and can be On any of the menu items that enable you to
calibrated for 0 dBm (into 500 or 600 Ω) modify the frequency, the unit’s output frequency
Ÿ Three variations of sine wave output full scale
will follow the changes as you make them with the
§ Simple sine wave – one frequency,
continuous output Ÿ Fits in a 5” × 3.15” × 1.4” box, QE. You can also use the joystick to modify the
§ Sine wave sweep making it easily portable (this size frequency but the change will not occur until you
w Linear sweep: user selects step box does not have room for batteries)
press Enter. The joystick can be used to move the
size
w Octave sweep: user selects full Ÿ Can be powered by 3 AA cells— cursor to different digits of the frequency. Both the
octave or fractional values of alkaline battery life should be more QE and the joystick will update the frequency based
1/2, 1/3, or 1/4 than 40 hours if neither the LEDs nor
on the digit selected. For instance, if the joystick is
w User selects lower and upper the backlight are used
frequencies, step duration, one used to move the cursor to the hundreds digit, the
marker frequency may be set Ÿ Uses a PIC18F25K20 microcontroller frequency will be modified in 100-Hz steps.
§ Frequency list of up to 32 discrete and an AD9834 DDS device
The menu system is hierarchical but most entries

46 | June 2015 | audioxpress.com

 Menu Selection List
Ÿ Single frequency Ÿ Sweep Ÿ List
§ Frequency § Low frequency § Count
§ High frequency § Duration
do not have any sub menus other than entering Ÿ Save § Step size § Frequencies
§ Main menu w Fixed frequency § Go
the frequency value (see the Menu Selections list w Octave
§ Single frequency
for the options). § Sweep w Octave/2
Both the Sweep (fixed step size only) and List § List w Octave/3
w Octave/4
selections have additional manual stepping features. § Step duration
If at any time during the Sweep or List output you § Marker frequency
press Enter, the output frequency will stop advancing § Go
and the program will wait for another switch
input. Pressing Enter again will cause the output Initially, I did not have a reset circuit for the
to continue from where it was paused. Moving master clear reset (MCLR) pin of the PIC. It worked
the joystick to the right will cause the next higher fine when using the debugger. However, when I
frequency to be output. Moving the joystick to the attempted to use it stand-alone, it would not always
left will cause the previous (lower) frequency to be properly start. I then implemented the MCLR circuit
output. The Sweep/List operation may be terminated as recommended in the processor manual, which
by activating any joystick position while the normal then enabled the correct operation. When operating
frequency advancing is in progress. the generator without the debugger, a jumper plug
must be installed between pins 1 and 2 of H105 to
Hardware connect the reset circuit to the MCLR.
Figure 1 shows the connections to the PIC. One If you want to customize the program, you will
of the most important design criteria, which must need to remove the jumper to connect the debugger
be decided at the beginning of any microprocessor cable to pins 2–7. The pinout of H105 (pins 2–7) match
project, is to assign each processor pin to its that of the PICkit-3. I made an extension adapter for
required function. It is also a good idea to keep the my PICkit-3 that enabled me to plug into H105 as well
signals grouped by function, as much as possible, as into a solderless breadboard. You may want to
utilizing the features of PIC’s different I/O pins. I make your adapter with a seven-pin connector, with
put the QE outputs on RB0 and RB1 to use the RB0’s pin 1 blank, to help ensure you do not plug it in wrong!
external interrupt capability. I had to use RC3–RC6, H101 is a right angle header. It should be
although RC4 is not needed, as the serial peripheral mounted such that its pins face the board’s center
interface (SPI) port for communicating with the DDS. and be over the regulator IC. The reason for this is
I also wanted to keep the debug pins (RB6 and to keep the unit as thin as possible. If the header
RB7) dedicated to that function. That left only the was vertically mounted, its mating connector would
Port A pins for use as the LCD interface. I decided add about 0.5” to the unit’s height.
to use the upper four bits for two reasons: The LCD Looking carefully at Figure 1, you will notice that
uses bits 4–7 for its 4-bit interface and the sequence
for sending a byte to the LCD via its 4-bit interface
sends the upper four bits first. You can see from
Figure 1 that RC4 is the processor’s only unused pin.
The PIC I selected, the 18F25K20, is a 3.3-V
part, which I chose because I happened to have
a 3.3-V LCD in my parts bin. It is also in the same
family as the processor in the PICkit3. If you would
rather use 5-V parts, there is an equivalent PIC, the
18F25K22, which should work fine on the PCB. The
code should not have to change based on what I
have seen in the 18F25K22’s datasheet.
The parts required for a 5-V system are listed as
options on the bill of materials (BOM) found in the
Supplementary Materials section of the audioXpress
website. The remaining circuits will work with a 5-V
supply. I used a 5-V oscillator, which works just fine
with 3.3 V. However, the PCB is designed to handle
surface-mount device (SMD) oscillators as well as
the four-pin DIP oscillator I used. There are two
SMD oscillators listed as options in the BOM, one
is 5 V the other is 3.3 V. Figure 1: All the connections to PIC are detailed.

audioxpress.com | June 2015 | 47

 ax You Can DIY!

Figure 2: The LCD circuit is LEDs with 0.1” (2.54 mm) lead spacing.
pretty straightforward.
The LCD circuit is quite straightforward (see
Figure 2). R201 is used to adjust the contrast. It
can usually be eliminated by simply connecting
its center to ground. H201 is used to connect the
LED backlight to the power source. I implemented
separate connections for the backlight since my unit
draws about 150 mA and I did not want that current
flowing through the PCB ground. However, the newer
units, specified in the BOM, draw considerably less:
15 mA.
Since the current is less, you may want to simply
install a jumper from the LCD end of R205 directly
there are two grounds—an analog and a digital. It is to the output of the regulator. If you still want an
About the Author probably not necessary for this circuit but it should external resistor then its value may be different,
J. Lawrence Cicchinelli, be considered when designing mixed signal circuits. depending on the main power supply you use. The
o f C a m d e n t o n , M O, H102 is not a physical header—it is there just for its LCD has a built-in resistor to limit the current to
received his BSEE pads, which are used to connect the two grounds the LEDs based on the normal LCD power supply.
from Drexel University
together. You must install a jumper wire between You can calculate the value for R205 as follows:
Institute of Technology
in 1969 and his MSES the two or the circuit will not properly work.
f ro m Pe n n s y l v a n i a Neither H103 nor H104 need to be installed. The R205 = (VPOWER – VDD)/ILED,
St ate University in pads for H103 may be used to view the SPI signals
1981. He worked for for U302—the DDS chip. The H104 location is the only if you want full brightness. On my system, I decided
Ford Motor Company
provision on the board for the Marker output. There to use 100 mA for the LEDs and a 9-V wall wart to
f ro m 1 9 6 7 t o 2 0 0 0 ,
including 30 years is also an optional resistor that enables LED103 to power the unit so my resistor is 56 Ω: (9-3.3)/0.1.
designing and building turn on whenever the Marker frequency is being The DDS circuit is based on the AD9834 (see
test equipment for output. If you want this option, you should install Figure 3). I have used this circuit two previous times
automotive electronics. R107 instead of R103. so I already knew how to program it. It uses a 28-
He was the technical
If you plan to primarily use battery power, you bit accumulator so I would liked to have used an
support manager for
Digi International, might want to eliminate the LEDs, including the oscillator frequency of 26.8435456 MHz (228/10) to
R a b b i t B r a n d , f ro m LCD’s LED backlight to save the batteries. If you do achieve 0.1-Hz resolution simply by multiplying the
2000 to 2012 and wrote want the LEDs, you might want to search for more desired frequency in hertz by 10. That would have
the book Assembly efficient ones that will not require a lot of current. given me the value to directly program into the
Language Essentials
There are some high-efficiency T-1 LEDs available frequency registers. However, getting an oscillator
published by Circuit
Cellar in 2011. that are reasonably bright with only 2 mA. If you do of that frequency requires a custom order, which is
use lower current LEDs, you will want to change the quite a bit more expensive than I wanted. I settled
values for R101–R103. They currently supply about on a standard 25-MHz oscillator, which is very
6 mA to each LED. The circuit board is designed for inexpensive, but it forced me to do some creative

Figure 3: The DDS circuit is based on the AD9834, which is a low-power DDS device.

48 | June 2015 | audioxpress.com

DESIGNED
FOR MORE
SENSITIVITY
POWER HANDLING
AFFORDABILITY
We used our decades of award-winning design and manufacturing
expertise to create a line of professional transducers that delivers
superior sound – for a lot less. Our column transducers feature a
compact, wide bandwidth design with FEA-optimized geometry
and our patented Penta Cut cone technology for better frequency
response. Each component has been thoughtfully engineered to
withstand the rigors of professional applications.

Visit us at InfoComm in Booth 258.

See all of our TC-Series full range and
professional transducers at Tymphany.com/TC.

TC-SERIES
2.5″ FULL RANGE

©2015 Tymphany
 ax You Can DIY!

With the resistor values as shown on the schematic,

the theoretical peak-to-peak output voltage range
is 0.288 to 0.864 VPP, although the compliance of
the output for DAC is specified as 0.8 V.
The system is somewhat limited as to its output
voltage since it uses a 3.3-V supply. The offset circuit
on U302.2 is needed for two reasons: (1) the circuit
is an inverter configuration but cannot have the
output attempt to go negative and (2) even a rail-
to-rail op-amp does not always include 0 V in its
output range.
The equation for the output voltage of U303.2 is:

VOUT = VOFFSET × (1 + R309/R308) – VIN × R309/R308

Figure 4: The tone control uses a single potentiometer, which enables adjustment of the where: VOFFSET = 0.503 V and VIN is the voltage
frequency response from base boost/treble cut through to treble boost/base cut.
at the arm of R310. The equation becomes VOUT =
3.02 – 5× VIN after inserting all the values.
math to get the correct value to program. Keep in mind that this is the DC equation and
R303 is used to calibrate the full-scale output that the AC output is 5 × VIN as long as VIN does not
current from the DDS IC so that with R310 fully exceed 0.6 V. If R303 is adjusted for 2 VPP output
clockwise, the output voltage is the desired from U303.1 with R310 set for maximum, then the
maximum value: 2 VPP is 0 dBm into 500 Ω and output of U302 will be 0.4 VPP.
2.19 VPP is 0 dBm into 600 Ω. The DDS’s output DAC The output circuit is unbalanced and uses
delivers a current with the formula: capacitive coupling. I would have preferred using
a transformer but could not find a low-cost one
IOUT(full scale) = 18 × VREF/R SET with the desired bandwidth.

where VREF is fixed at 1.2 V. Since the output resistor Software

is 200 Ω, the formula becomes: The program is entirely written in Assembly
Language. I purchased a PICkit-3 specifically for
VOUT = 200 × IOUT = 200 × 18 × 1.2/ RSET = 4,320/ RSET. this project and used MPLab X for the development
environment. To help me remember what functions I
wrote and what parameters they require, I created a
text file that basically contains the headers for each
function. Since I am a firm believer in documenting
code, all the lines of code are commented as are the
functions. The complete MPLab project is available
on my website as well as from audioXpress (see
Resources).
Most of the arithmetic functions are specifically
written for this project. Since the design upper limit
frequency is 100 kHz with 0.1-Hz resolution, the
functions only need to handle a maximum of 20 bits
to program the DDS chip. This enables me to limit
the functions to handling only three-byte values.
The DDS IC’s output frequency is:

(FClock /228) × N

where N is the value programmed into its 28-bit

accumulator. Since I am using a 25-MHz oscillator,
the equation becomes: Frequency = 0.0931322 × N.
Since I know what frequency I want, solving the
Photo 2: The frequency response is shown with the control roughly centered. equation for N results in: N = 10.737418 × Frequency.

50 | June 2015 | audioxpress.com

The first thing the algorithm does is multiply
the desired frequency by 10. That result is then
multiplied by 1.073730, which you can see is close
to the desired value. The method I use to generate
the fractional multiplier is to add values of the
frequency × 10 divided by appropriate powers of
two as follows, let Frequency × 10 be represented
by F × 10 so, N = F × 10 + F × 10/16 + F × 10/128
+ F× 10/512 + F × 10/1,024 + F × 10/2,048).
To get closer than this to the “real” value, the
program would have to execute six more shifts and
an addition. I did not think the extra processing
was worth the minor improvement.
I wrote two functions to help me with the
calculation: 3-byte shift right and 3-byte add. To
conserve memory, the 3-byte shift is done “in place”
Photo 3: Here the frequency
while the 3-byte add updates an accumulator but included with the order. If you use this service, be response is shown with the
does not modify the shifted value. This enables me sure to include whatever sales taxes are applicable. control in the treble boost/
to use the shifted value for both the add operation There are only two parts in the BOM that are not base cut position.
as well as the succeeding shift operations. There from Digi-Key—the 5-V LCD and the knobs. You might
are other math functions that handle 3-byte values: notice that I have specified mostly 1% resistors. I
multiply by 10, shift left, and subtract. have found that 1% and 5% SMD resistors are the
I also needed a divide function that would handle same price so I always get the 1% units.
3-byte values. I was all set to write one myself but I did have a little difficulty implementing the
I found a very fast function on the Internet that joystick due to its shaft, which is small. I found an
handles a 24-bit dividend and a 16-bit divisor—just aluminum standoff in my stock that fit nicely over
what I needed! The source for the function is noted the shaft. I filled one end of the standoff with glue
in the Project Files found in the Supplementary and put the joystick shaft into it. It works nicely
Materials section of the audioXpress website. I only as a shaft extension. The part listed on the BOM
had to make a few minor changes to accommodate is similar. It should work, but I have not tried it.
my specific memory usage. I obtained the knobs from a friend. Each of
T he Microchip website cont ains all the the three controls has a 6-mm shaft, including the
documentation on the PIC processors as well as extension on the joystick. However, I did have to
MPLab X, which is a free download. There is a link on modify the joystick knob. I ground some of the
the homepage under Design Support for the official bottom off so it would not hit the top surface of
Forum which has lots of help for all users. To find the box as the switch is moved around. The knob
the divide function previously mentioned, I entered listed on the BOM is similar—at least its dimensions
“PIC integer divide” into my search engine and I are what is needed.
was presented with several pages of possibilities. The socket (BOM1) and header (BOM2) listed for
the LCD had to be cut to the correct size (8 pins each).
Construction Notes You will need two of each. I mounted the headers on
I have been using DipTrace for all my schematics
and PCB designs for several years. There is a free
version available from www.diptrace.com, which is Project Files
To download additional material and files, visit http://audioxpress.com/page/
fully functional. All of the design files are available
audioXpress-Supplementary-Material.html.
from my website (www.qsl.net/k3pto) as well
as from the Supplementary Materials section of
Resources
audioXpress (www.audioxpress.com). DipTrace software, www.diptrace.com.
The BOM for the system includes all the parts
for the circuit board—except for some mounting K3PTO Published Articles, www.qsl.net/k3pto.
hardware—and the vendor part numbers. The power
connector and the switch are parts I found in my Source
MPLAB PICkit 3 and MPLab X
stock and can be whatever you want to use. Microchip Technology, Inc. | www.microchip.com
I purchase most of my parts from Digi-Key
Electronics since it has free shipping if a check is

audioxpress.com | June 2015 | 51

 ax You Can DIY!

the LCD and the socket on the PCB, although it would might seem a bit large but it must enable the shaft
work just as well if you reverse them. to move from side to side and top to bottom. I
Each time you cut a socket to size you will lose generally drill my holes a little bit small and then
one pin. Cutting the header does not sacrifice any enlarge them using a reamer until I get them to
pins. One thing I like to do with all my connectors is the right size.
to “paint” them with a permanent marker so I can I soldered the LEDs onto the PCB last. To get the
easily see pin 1. I have a silver marker that shows correct height, I mounted the PCB into the box with
up quite nicely on the black body of the headers. the LED leads in their holes. I pushed the LEDs up
I created a template for drilling the top of the from the bottom of the PCB, until they protruded
case, based on the component centers on the PCB. through the box by the amount I wanted. Then, I
The file should be printed using DipTrace since it soldered them into place.
will print 1:1. There is also a .jpg file, which can be
used as a reference. I put the dimensions on it so Usage
the size can be verified after printing. The cutout One simple use of the generator is to characterize
for the LCD is also indicated using dimensions. audio filters. I built a simple tone control circuit for
Although there are two mounting holes in the PCB demonstration purposes (see Figure 4). The tone
for the LCD, I did not use them. The two 8-pin control uses a single potentiometer, which allows
sockets hold it in place. The hole for the joystick adjustment of the frequency response from base

Direct Digital Synthesis Primer

Direct Digital Synthesis (DDS) is a method of developing a exactly 0.1 Hz and made the calculations much easier.
programmable frequency using primarily digital techniques. The operation of the Phase Register is such that the value of
The only analog part of a DDS system is a digital-to-analog M is added to its content with every FC clock. When it overflows
convertor (DAC) and whatever filters might follow it. One of a its range, it is reset to 0. The value of N for most DDS ICs
DDS’s primary features is that it can develop frequencies over ranges from 24 to 48 bits.
a large range with a small value of resolution. The DDS device Also, and quite importantly, the data to the Lookup Table
used in this article could develop frequencies from 0.1 Hz to is truncated to the number of bits required by the DAC. Also,
more than 10 MHz with 0.1-Hz resolution. since a sine wave is symmetrical in its four quarters, the most
Figure 1 is a simplified block diagram of a DDS system, significant two bits are used to control some mapping logic
where FC is the main oscillator, or reference clock, which must to create the full sine wave from a Lookup Table containing
be at least twice the highest desired output frequency. This is the sine values for 0° to 90°. With a 10 bit DAC, the Lookup
due to the Nyquist Criteria, which basically states that a signal Table would be driven by bits 16-25 for a DDS IC with a 28-bit
must be sampled at least at twice its frequency in order to Phase Register.
reproduce its frequency. Another method of generating a programmable frequency
M is the tuning word, which is the mechanism for is the Phase-Locked Loop (PLL), which has been around for
determining/controlling the output frequency. FOUT is the output quite a few years and is still a viable method for very high
frequency and is equal to M × FC /2N. frequencies. One of the main differences between a PLL and a
You can easily see from the formula that if FC = 2N, the output DDS is that the DDS will generate frequencies below its main
frequency will be the value of M in hertz, which is why I wanted oscillator while a PLL generates higher frequencies. Also, a
an FC of 26.843545 MHz for the 28-bit Phase Register (228 = PLL utilizes an analog oscillator that is usually controlled by an
268,435,456). This would have given the system a resolution of analog voltage whereas the DDS is completely digital except for
the DAC, which does not enter into the control loop.
An advantage of the PLL is that its output can be very
“clean” (i.e., there should be few harmonics or other noise/
distortion components to the signal in a properly designed
system). However, the DDS will have noise components that
must be filtered out. The main advantage of a DDS is that it
enables large frequency changes with high speed and excellent
resolution. It is also possible to control a DDS signal’s phase
Figure 1: This is a simplified block diagram of a Direct Digital
Synthesis (DDS) system.
so that you can have multiple outputs with known phase
relationships among them.

52 | June 2015 | audioxpress.com

boost/treble cut through to treble boost/base cut.
The generator is set to its sweep mode with the
following parameters:

• Start frequency = 100 Hz

• Stop frequency = 55 kHz
• Step size = 1 octave
• Duration = 100 ms
• Marker = 100 Hz

With the above settings, the output frequencies,

in hertz, will be: 100, 200, 400, 800, 1,600, 3,200,
6,400, 12,800, 25,600, and 51,200 Hz. The only reason
I set the upper frequency so high is that I wanted 10
frequencies to have a “nice” oscilloscope display. With
the oscilloscope set to 100 msec/div, each division
represents a single frequency. flat across the control’s entire range. Photo 3 shows Photo 4: The frequency
Photo 2 shows the frequency response with the the frequency response with the control in the treble response is shown in
control roughly centered. Trace 1 is the Marker, which boost/base cut position. Photo 4 shows it in the base the base boost/treble cut
position.
is set to trigger when the frequency is 100 Hz. The boost/treble cut position. ax
slightly reduced signal at the high end is probably due
to the circuit’s simplicity.
I did verify that the output of the generator is

EMPOWER THE MUSIC

Read.
Design.
Build.
ENJOY!

With this how-to loudspeaker book, you will be

able to crank up the volume on a first-rate system
that you designed and built yourself.

cc-webshop.com
audioxpress.com | June 2015 | 53
 ax You Can DIY!

Build a Single-Ended
Guitar Tube Amplifier

Photo 1: This small, single-

ended guitar tube amplifier is
great for studio use.

Music, especially rock, jazz, blues, and pop owes much to the electric guitar and to
vacuum tube amplifiers. I enjoy building simple tube amplifier designs because I believe
simple circuits work better. Simple circuits are more effective, especially when the electric
signals carry music information, because the signal distortion is easily handled.

By
Costas Sarris
(Greece)

When a guitarist friend urged me to build a The Circuit

guitar tube amplifier, we agreed that the amplifier The music information generated as an electric
design should be a simple, low-wattage single- signal by an electric guitar pickup is weak and needs
ended amplifier—built much like a Swiss army amplification. Guitar pickups have different output
knife—for everyday use. Inspired by late 1950s voltages depending on the model. Low-output models
Fender “Princeton” and “Champ” amplifier designs, tend to produce a “clean” sound, while high-output
I decided to construct a small but efficient single- models tend to overdrive amplifiers and produce a
ended head amplifier, that any DIYer could easily “dirty and aggressive sound.” The output voltage of
build (see Photo 1). most pickups varies between 100 mV and 1 VRMS.
This small amplifier is a medium/high gained This amplifier uses the high gain 12AX7/ECC83
head, meant to be played at room or studio levels. double triode vacuum tube for the preamp stage, and
Although it can get pretty loud, it can also be the 6L6GC beam power tube for the output stage in
played with smooth bass and “quiet drums” on a single-ended Class-A design.
small stages. Class-A amps are voltage amplifiers. In a Class-A
Because of its warm sound, the amplifier isn’t amplifier the output voltage’s wave shape is the same
made for any kind of hard metal music. On the other with the shape of the signal voltage applied to the grid.
hand it is ideal for blues, rock, jazz, or funk music. In other words, the power tube runs at full power all
The essence of this amplifier is simplicity. That’s the time and that affects the way the output tube
what makes it different (see Photo 2). distorts. A Class-A power amplifier typically sounds

54 | June 2015 | audioxpress.com

warm and natural. When it tends to clip, it sounds
dynamic and aggressive. That’s the reason these
amplifiers are desirable for many guitarists.

Preamplifier Circuit Design

The preamp stage uses a high gain 12AX7/ECC83
dual triode vacuum tube, and operates as a grounded
cathode amplifier. The output is taken from the second
half of the 12AX7 triode plate to the power tube’s grid,
using a high-quality coupling capacitor 0.1 µF. The
driver tube is cathode biased with a 1.5-kΩ resistor
(R8, R2 shown in Figure 1), and a 200-V plate voltage
(both stages).
The preamplifier’s first half stage is a fully bypassed
cathode bias circuit. The input grid-stopper resistor
R12 is 33 kΩ. To bypass the cathode resistor in the
first stage (the first half triode), I used a 25-µF/25-V
audio electrolytic capacitor (C2 shown in Figure 1).
The gain potentiometer (1 MΩ) acts as a variable
grid resistor to the grid of the second half of the 12AX7
preamp tube. The tone control section consists of a
simple “tone boost control” circuit. C6 is the treble-
shunt capacitor, so P2 (the tone control pot) affects
both the ”tone cut” and the boost. The amount of boost Photo 2: This vacuum tube amplifier generates a warm sound.

Loudspeaker Transducers

Danish R&D Staff

and Danish heritage

www. WAVECOR.com

audioxpress.com | June 2015 | 55

 ax You Can DIY!

Figure 1: The amplifier’s

design uses 12AX7/ECC83
dual triode vacuum tubes.

About the Author

Costas Sarris earned
his BS in Business
Computers from North
College Thessaloniki
i n 1 9 9 0 . To d a y, h e
works as a Medical
Systems Engineer
for BioLab Fostieris,
a leading medical
diagnostics company
in Greece. He has been
constructing handmade
tube amplifiers since depends on the gain pot’s setting. Tweaking the gain, about a 20-V signal amplitude to be driven to full
1992 as a hobby and you can adjust the amount of distortion you want. power. In this circuit, the grid resistor is a variable
decided to share his The second half of the triode stage can be with resistor (1-MΩ master volume pot). So if we use a small
audio tube amplifier or without a cathode bypass capacitor. Decreasing size (e.g., 1,000 pF) cathode bypass capacitor in the
passion with the DIY
the cathode bypass capacitor’s size, improves the
audio community. He
lives in Thessaloniki, amplifier’s transient response.
Greece, with his wife While the power tube cathode resistor is fully
and daughter. When bypassed (C1, 25 µF/25 V), the power stage needs
he is not constructing
tube amps or listening
to music, he mountain
bikes.

Photo 3: For this amplifier,

I chose a custom-made,
E-shaped, high-quality
single-ended output Photo 4: I used a filtered IEC power input connector for
transformer. the AC input in the power supply.

56 | June 2015 | audioxpress.com

 Photo 5: The wood casing is made
from handcrafted 16-mm thick birch
plywood.

preamplifier’s second stage, the amplifier’s Output Transformer

input sensitivity—with the master volume pot One of the most critical components in
fully open (1 MΩ)—is only 15 mV. That means the sound path is the output transformer.
we only need a 15-mV input to overdrive Therefore, you should never compromise on
the power stage! Using a high-output guitar the output transformer’s quality.
pickup, we can easily overdrive the amplifier For this amplifier, I chose a custom-made,
to produce the desired “aggressive sound.” “E” shaped, high-quality single-ended output
Increasing or decreasing the cathode bypass transformer (see Photo 3). The resonance-
capacitance in the second half of the driver free frequency range for this transformer
stage, we can improve the tone response, is much better than 80 Hz to 20 kHz
according to our individual taste. (–3 dB). The transformer has two secondary
windings, 4 Ω and 8 Ω. I also recommend
Power Amplifier using Hammond Manufacturing’s 125ESE or
In this project, the 6L6GC power tube is 125FSE output transformers.
running at a 58-mA plate current when no
signal is applied, and a 350-V plate voltage. The Power Supply
The output transformer’s primary impedance When designing a single-ended tube
4.5 kΩ. The power tube cathode resistor is amplif ier circuit, the power supply
fully bypassed by a 25-µf/25-V capacitor (C1). section must be carefully considered.
The input grid-stopper resistor is 1 kΩ (R1). For this proje c t, my cu s tom-ma de
The voltage across the cathode resistor is power transformer used the following
measured at 19.5 V. windings:

Photo 6: I soldered all the parts to the soldering tag board prior to the assembly.

audioxpress.com | June 2015 | 57

 ax You Can DIY!

• The primary winding was a 230 VAC (110 VAC in

the US)
• The secondary windings were 6.3 V/1.5 A
filaments for 12AX7 and 6L6 power tube and
260 V-0-260 V/120 mA

Hammond Manufacturing’s 269A X power

transformer is a good substitute and can be easily
found. The rectifier circuit uses a single section “pi”
filter or a “capacitor-input filter.” Both the filter
capacitor C11 and the smoothing capacitors C9
and C10 are 47-µF/500-V electrolytics. The rectifier
bridge circuit in the power supply section consists
of two 1N5407 general purpose rectifiers. Filaments
are powered with 6.3 VAC for driving and power
tubes. For the AC input in the power supply, I used
a filtered IEC power input connector (see Photo 4).

Assembling the Amplifier

For the amplifier’s construction, I used a ready-
made 1.2-mm thick aluminum chassis. The dimensions
Photo 7: To complete the wiring, I connected all the parts directly from tube sockets and
were 25 cm × 15.5 cm × 4.5 cm (W × D × H).
input/output jacks to the tag board.
I made the wood casing from handcrafted 16-mm
thick birch plywood (see Photo 5). Then, I covered
the wood casing in leather. Tolex or textured vinyl
can also be used.
Before the final assembly, I soldered all the parts
to the soldering tag board (see Photo 6). All the pots
are Alpha 24-mm full-shaft 1-M linear.
I installed all the basic components on the chassis
(the tube sockets, the output transformer, the power
transformer, the pots, the input, and the speaker
jacks, etc.). Then, I finished the basic wiring starting

Parts List
Capacitors
C1 . . . . . . . . . . . 25 µF/25 V
C2 . . . . . . . . . . . 25 µF/25 V
C3 . . . . . . . . . . . 1 nF/25 V
C4 . . . . . . . . . . . 0.1 µF/500 V
C5 . . . . . . . . . . . 0.01 µF/500 V
C6 . . . . . . . . . . . 1 nF/500 V
C7 . . . . . . . . . . . 0.01 µF/500 V
C9, C10, C11 . . . 47 µF/500 V

Fuse
F1 . . . . . . . . . . . 1.5 A/ 250 V

Potentiometers
P1, P2, P3 . . . . . 1 M/alpha 24-mm linear potentiometer

Rectifiers
D1, D2 . . . . . . . . 1N5407 rectifiers

Tubes
6L6GC . . . . . . . . power vacuum tube
12AX7/ECC83 . . . preamp driving vacuum tube

58 | June 2015 | audioxpress.com

 7052PH
Phantom Powered
Measurement Mic System
NOW 4Hz to 25+kHz
<16dBA >135dBSPL
IEC61094-4
Compliant 1dB/div
30 kHz

Photo 8: All the tube sockets were made of high-quality porcelain.

with the filaments. I used 1.2-mm wire For our tests, we used a P12R 12” Alnico
for the ground plane line, and for all the speaker.
critical paths. All the soldering underneath
the chassis, is hand-done, point-to-point In Practical Application
wiring using a soldering tag board to place “Costas Sarris was kind enough to let Anechoic
the parts. me try one of his little wonders, and I’m
All the power supply parts soldered so happy he did that!” said Zaikos. “I’m
Free Field
in a separately tag board, installed in a working musician, I play tough and Response
the left side of the chassis underneath hard, electric blues. Guitar sound needs Included
the power transformer. To complete the harmonics, tone, bite, and character. His
wiring, I connected all the parts directly amp gave me the chance to travel from
from the tube sockets and input/output sweet and bold single notes to cutting
jacks to the tag board (see Photo 7). and exploring double stops and all points
All the tube sockets were made of high- between, really nice! To make it short,
quality porcelain (see Photo 8). For the when you play the blues on the road, Rugged
final step, I mounted the chassis to the all you need is an amp to “hear” you, STAINLESS
wood casing. to feel your touch and express your
Body
deepest inner self...and that ain’t easy.
Test Impressions Costas’ little amp made me feel like I Operational
Two guitarists tested the amplifier. One was playing through an old trusty sonic <-20 to >70C
is Elias Zaikos, a legendary guitarist and box. I couldn’t ask for more. Clear and Storage
founder of Blues Wire (www.blueswire. gently or nasty and growling, this was a <-30 to >85C
gr), and the architect of blues in Greek fantastic experience, Costas, thank you
blues scene. The other guitarist and a for sharing your work!” Phantom
good friend is George Liagas, owner and So plug in and rock! Cheers! <18Vdc
designer of Warlord custom pedals (www. to
guitarpedals.eu). He is also the one who Author’s Note: I would like to acknowledge >56Vdc
urged me to build this amplifier. George Liagas, founder and designer of
Both guitarists agreed that with 10 W Warlord custom pedals for his infinite help Removable
of output power, this amplifier rocks with in testing and fine tuning the amp and Elias
amazing tone, and sounds superb when Zaikos for his effort testing and tweaking Capsule
combined with a good-quality speaker. the amplifier. ax Titanium
Diaphragm
Resources
M. Blencowe, Designing Valve Preamps for Guitar and Bass, Wem Publishing, 2009.

R. Honeycutt, “Tone Controls,” audioXpress, August 2013.

M. Jones, Valve Amplifiers, Newnes, 1999.

RCA Receiving Tube Manual, Technical series RC-19, RCA, 1959. www.acopacific.com

audioxpress.com | June 2015 | 59

 ax Hollow-State Electronics
Finding Information
on Vacuum Tubes

Photo 1: Manufacturers’ databooks

This article attempts to unravel some are valuable references for
of the challenges associated with electronics hobbyists.

finding information on hollow-state

components and how to determine
what the data actually means. Even
with the vast resources of the Internet,
finding useful data can still be difficult.

By
Richard Honeycutt
(United States)

The late 1950s and early 1960s was a challenging Interpreting the Data
time to be an electronics hobbyist. The availability The second part of a hobbyist’s problem —
of information on hollow-state components was understanding the data—often comes down to
increasingly limited to older books and a few recognizing the necessary and achievable degree of
magazines, in anticipation of the widely-heralded accuracy in the circuit-design process. For example,
death of vacuum-tube technology. Most of the we often give the equation for the gain of a triode
books on solid-state components were textbooks: circuit as:
manufacturers’ data was hard to come by, and
µRl
understanding what the component data meant— AV =
when you could actually find data—was not a rP + Rl
straightforward process.
Fast-forwarding to today, we find that in some where μ = the amplification factor of the triode, rP
respects, the hobbyist has an easier time. The = the value of the triode’s plate resistance, and R l
Internet provides a way to access data on most = the AC load resistance connected to the triode
components, be they solid- or hollow-state. (To plate (i.e., the plate resistor’s value paralleled with
borrow from the popular Mark Twain misquote, the impedance of whatever the amplifier stage’s
reports of the death of hollow-state technology plate is feeding).
turned out to be greatly exaggerated!) Yet, often it is The typical assumption most of us would
more convenient to have a databook at one’s elbow make upon seeing this equation is that just as the
than to have to shuffle among multiple webpages resistance of a carbon-film resistor is a constant
for specifications on several components. And, the value, so are the μ and rP of the triode. Is this true?
problem of understanding the data remains. No. Does it matter that these values depend upon

60 | June 2015 | audioxpress.com

 Figure 1: This is the first
page of the General Electric
(GE) datasheet for the
12AX7A.

the tube’s operating conditions? Sometimes. Without These books are now available from eBay,
getting into the physics of a tube’s construction Amazon, and perhaps other Internet sellers, and
and what elements affect the μ and r P, let’s see in .pdf form from a several websites. Individual tube
what information we can find, and then figure out datasheets can also be downloaded from frank@
what it means. pocnet.net and 11 mirror sites. As we will see,
different manufacturers include different forms of
Understanding the Datasheet tube data in their datasheets, and tube manuals
The most common information source about a may contain more than just datasheets.
particular vacuum tube’s operating characteristics Figure 1 shows the first page of the General
is a datasheet. For many tubes, these datasheets Electric (GE) datasheet for the 12AX7A dual triode.
have been compiled into “tube manuals,” as RCA At the top is the tube’s general description and
called them, or “technical manuals,” as Sylvania its intended applications. Below this description
called them (see Photo 1). are electrical and mechanical characteristics.

audioxpress.com | June 2015 | 61

 ax Hollow-State Electronics
The electrical characteristics include the heater (one side of the heater supply connected to pin 4
(filament) voltage and current both for the full and pin 5, and the other connected to pin 9). The
heater—pin 4 and pin 5—connected across a 12.6-V heater’s electrical data is important for specifying
supply (“Series”), and for the two halves of the the power transformer’s filament winding ratings.
heater connected in parallel across a 6.3-V supply The other class of electrical characteristics is the
interelectrode (“parasitic”) capacitances. These
capacitances affect the frequency response of the
amplifiers built using these tubes.
The mechanical characteristics tell the designer
what kind of socket the tube needs (the “Base”), and
how much physical space to allow for each tube (the
“Envelope”). The envelope number, T-6 ½, may be
keyed to a dimensioned illustration in another part
of the tube manual. In this case, the dimensioned
outline appears at the lower left of page 1 of the
datasheet.
The data section entitled “Maximum Ratings”
is pretty much self-explanatory. These are the
ratings that should not be exceeded when the tube
is operating. Unlike solid-state devices, vacuum
tubes will survive short periods during which the
maximum ratings are exceeded, although the life of
the tubes may be diminished. Some datasheets use
the term “design-maximum values” to indicate these
specifications. They may even specify maximum
instantaneous values, which, if exceeded, may result
in the tube’s catastrophic failure.
The only maximum value that may need
clarification is the plate dissipation. In a Class-A
amplifier stage, the average value of plate current
remains constant (often called “IPQ,” for quiescent
plate current). The product of IPQ × VP equals the
tube’s plate dissipation. This is not the same as
the power output of the tube stage. In fact, for a
Class-A stage, the sine-wave power output cannot be
greater than one-third of the plate dissipation. Plate
dissipation causes heating of the tube’s plate. If the
plate dissipation becomes too great, the tube can
“red-plate,” meaning that the plate can be heated
red-hot, with the possible result of the plate and/
or its supporting structures sagging and damaging
Figure 2: This is the second page of the General Electric (GE) 12AX7A datasheet. or destroying the tube.
The information found at the bottom section of
Figure 1 is self-explanatory. Figure 2 shows the
Resources second page of the GE 12AX7A datasheet. The
“General Electric 12AX7-A Twin Triode” datasheet, www.mif.pg.gda.pl/
homepages/frank/sheets/093/1/12AX7A.pdf. information shown on this page applies to the tube’s
typical operation in a circuit. At the top, two Class-A
“RCA 6AV6 Twin Triode—High mu Triode,” Electron Tube Division Radio amplifier examples are given: one with a 100-V plate
Corporation of America (RCA), 1959.
voltage and another with a 250-V plate voltage.
Sylvania Radio Tube Technical Manual, Sylvania Electrical Products, 1951, Notice that even though the control grid is biased
www.tubebooks.org/tubedata/Sylvania/1951/sylvania_1951.pdf. more negatively for the 250-V case, the IPQ is twice
“Tung-Sol 12AX7 Specifications,” Tung-Sol Electron Tubes, www.tungsol.com/ as large as for the 100-V case. The µ is the same
tungsol/specs/12ax7-tung-sol.pdf. in both instances, but the plate resistance is lower
with the higher plate voltage. The transconductance

62 | June 2015 | audioxpress.com

 The Easiest Way
to Design Custom
Front Panels &
Enclosures

You design it
to your specifications using
our FREE CAD software,
Front Panel Designer

We machine it
and ship to you a
professionally finished product,
no minimum quantity required

●
Cost effective prototypes
and production runs with
no setup charges
Figure 3: This is the third page of the General Electric (GE) 12AX7A datasheet.
● Powder-coated and
anodized finishes in
is also higher. nor a VCCS. But just as the stage voltage
various colors
A tube’s plate resistance (rP) is the ratio gain can be expressed as:
of the change in the voltage across the
µRl Select from aluminum,
AV =
●
tube (VAK) to the plate current. It is also
rP + Rl
sometimes called the “dynamic resistance” acrylic or provide your
it can also be expressed as:
of the tube. The transconductance (gm) is own material
AV = g m rp
a sort of gain. It is the ratio of the change
in plate current produced in response to The table shown at the bottom of ●
Standard lead time
a given change in control-grid-to-cathode Figure 2 shows the component values,
voltage. Thus, whereas the µ is the gain the gain, and the maximum output
in 5 days or express
when the triode is viewed as a voltage- voltage of one triode of a 12AX7A with manufacturing in 3 or
controlled voltage source (VCVS), the gm 10 different combinations of B+ voltages 1 days
is the gain when the triode is viewed as a and component values. The upper six
voltage-controlled current source (VCCS). cases represent cathode-biased circuits,
A true voltage source has zero output probably biased for minimum distortion
resistance, but a triode amplifier stage (although the datasheet does not say so).
has an output resistance approximately The lower four cases are grid-leak
equal to rP. A true current source has an biased. Some tube manuals (e.g., the
infinite output resistance, since the output RCA Receiving Tube Manual) include a FrontPanelExpress.com
current does not depend upon the load section after the datasheets in which
resistance connected to it. Thus, a triode similar design tables are included, each
amplifier stage is really neither a VCVS table applying to a variety of tube types,

audioxpress.com | June 2015 | 63

 ax Hollow-State Electronics
transfer characteristic curve (plate current plotted
vs. control-grid voltage). The parameter that is held
constant for each curve of the plate characteristic
is the control-grid voltage. Thus, the value of rP can
be determined by inverting the slope of the curve at
the desired grid-bias voltage. This is true because:
v P ∆VP
rP = =
iP ∆I P

and the slope of the curve is “rise/run,” or ΔIP/ΔVP.

The slope is taken at the Q-point, which is the point
representing zero-signal plate voltage and current.
The tube’s transconductance at any operating
point can be found as the slope of the transfer
characteristic curve at that operating point.
Figure 4 shows the GE datasheet’s fourth page
for the 12AX7A, which contains two graphs that
enable the designer to find the value of µ, r P, and
gm for any chosen value of plate current.

Datasheets Differ Based on the

Manufacturer
Different manufacturers include different
selections of data in the datasheet for each tube
type. For example, the RCA datasheet for the 12AX7A
refers the reader to the 6AV6 datasheet, which
does not include the transfer characteristic or the
graphs of µ, rP, and gm vs. plate current. The Sylvania
Technical Manual shows no curves at all for either
the 12AX7A or the 6AV6. The Tung-Sol datasheet
does not include the transfer characteristic, and
includes the graphs of µ, rP, and gm vs. plate current
in a different format than that used by GE.
At least half-a-dozen other manufacturers’
datasheets can be found online, each with its own
Figure 4: This is the fourth in lieu of including the table in the datasheet. individualities. Thus, in seeking vacuum-tube data,
page of the General Electric Since a triode of a 12AX7A is virtually identical designers have many options from which to choose.
(GE) 12AX7A datasheet. to a triode of many other tube types, the design
data for the 12AX7A also applies to the 3AV6, the Author’s Note: In Hollow-State Electronics article
4AV6, the 6AV6, the 6EU7, the 12AV6, and the 20EZ7, “Classic Tube Power Amplifier Circuits,” which ran
as well as the ECC83 (the European number for in the April 2015 issue of audioXpress, there were a
12AX7A), the 7025 (a low-hum/noise version of few errors. First, the phase splitters in the Heathkit
a 12AX7A), the 6681, and the 7729 (which are and Fisher amplifiers are of the split-load variety, not
industrial versions). long-tailed pairs. In both cases, the phase splitter is
If you calculate the gain using the component the second stage, and it is followed by a balanced
values given in the chart, and a µ of 100, you will (Class-A push-pull) driver stage. Second, the cathode
not obtain the voltage gain predicted by the chart. circuit of the Fisher has cathode feedback provided
Part of the reason is that µ varies with the triode’s by a tapped winding of the output transformer, not
operating conditions. There may be other reasons a tapped inductor. Third, the transformer winding
as well, but in my experience, the voltage gain and feeding the screens of the McIntosh actually provides
maximum output voltage predicted by these design positive feedback to bootstrap the final driver and
charts is only approximate. output screens. My thanks to reader Roger Modjeski
Figure 3 shows the plate characteristic curve of The Berkeley Hi Fi School (berkeleyhifischool.com)
(plate current plotted vs. plate voltage) and the for alerting me to these corrections. ax

64 | June 2015 | audioxpress.com

 Join the Largest
Premier Gathering for
Audio Enthusiasts!
Saturday July 11th
8:00 AM to 5:00 PM
Springboro, Ohio

4 Events, 1 Day Only!

Join thousands of audio enthusiasts at the
Parts Express Springboro, Ohio headquarters
for the lowest prices of the year. There will be
huge discounts on everything in our Retail Store
during the Tent Sale, as well as unbelievable
savings on factory buyouts, scratch & dent items,
and one of a kind samples. The Audio Swap
Meet will be back this year by popular demand,
giving you the opportunity to buy and swap
anything and everything audio! Experience the
high fidelity sound of creatively built speakers at
the Speaker Design Competition, and hear some
chest pounding audio systems at the Auto Sound
Challenge. Make your plans now to join us for a
day of giveaways, food, vendor displays,
and incredible deals!

Visit midwestaudiofest.com
for more event information and registration.

725 Pleasant Valley Drive

Springboro, OH 45066-1158
1-800-338-0531

N
75
W E

Exit #38 S

73
S. Pioneer Blvd.
Sharts Rd.

Pleasant Valley Dr.

Parts Express
 ax Industry Calendar
Here are a few places where you might find a copy of audioXpress or possibly meet one of our authors and staff members:

May 26–28, 2015 19 th AES Brazil Expo

Convention: May 25–28
Exhibition: May 26–28
Expo Center Norte Pavilhão Amarelo, Vila Guilheme, São Paul, Brazil
www.brasilexpo.com.br/en
The largest event of its kind in Latin America, AES Brazil is not exactly the usual Audio Engineering Society (AES) convention you find in the
US or Europe. The Brazilian event is much more business oriented and actually combines several exhibitors displaying video, lighting, and rigging
solutions, apart from professional audio, with a dominant focus on live sound and professional audio. The convention itself is a well-attended
event, attracting audio engineers and high-level professionals mainly from Brazil. Most of the sessions are only in Portuguese. Organized in
cooperation with Francal and the local AES section, the expected 75 exhibitors, occupying a combined area of 5,000 m2 will be the focus for
international visitors, even though most of the large exhibits are from the largest local distributors or local manufacturers. Harman, Yamaha,
Meyer Sound, and Proshows will promote workshops and practical training sessions.

May 29–31, 2015 The Home Entertainment (T.H.E.) Show Newport

The Hotel Irvine, Irvine, CA, US
http://theshownewport.com
The Home Entertainment (T.H.E.) Show has announced sweeping changes for T.H.E. Show Newport Beach. T.H.E. Show
Newport Beach will now be held at The Hotel Irvine in Irvine, CA. T.H.E. Show Las Vegas, usually scheduled for January
will be put on hiatus after 16 years. The promoters explain “all of T.H.E. Show’s resources and time will be spent making 2015 a game-changer
and a spectacular event.”
Visitors to T.H.E. Show Newport Beach can visit the Headphonium exhibitors, the Marketplace, vinyl, CD, and accessories vendors all in one loca-
tion—a giant 15,000-ft 2 Grand Ballroom in the center of The Hotel Irvine. Returning to the event in 2015 will be the gourmet food trucks, a giant
high-end auto show, live entertainment day and night, wine tastings, and many popular educational seminars orchestrated by the Los Angeles and
Orange County Audio Society (LAOCAS).

May 31–June 5, 2015 Euronoise 2015

MECC Congress Center, Maastricht, The Netherlands
www.euronoise2015.eu | Euroacoustics.org
Euronoise 2015, the 10th European Congress and Exposition on Noise Control Engineering, will be held in Maastricht,
in the heart of Europe where the first treaties leading to the creation of the European Union were signed. Acousticians and noise experts from
all over the Europe will gather for the event on noise control, and soundscape in Europe, organized by the European Acoustics Association (EAA).
Promoted this year by the Belgian and Dutch acoustical societies, ABAV and NAG, the Euronoise 2015 conference comprises keynote lectures,
tutorials, parallel technical sessions, an exhibition, and short courses. The program focuses on environmental noise control, acoustics, and
improvement of people’s quality of living. The organization confirms it has received more than 600 abstracts.
Euronoise 2015 will host an extended European Acoustics Exhibition, addressing all European providers of acoustic products, services, and
information. In this exhibition, companies, manufacturers, engineering service providers, research and administrative units, technical and scientific
organizations, and publishers will discuss the newest results in research and development, about equipment and instrumentation, methods and
software solutions and materials, standards, guidelines, and publications.

June 13–19, 2015 InfoComm 2015

Orange County Convention Center, Orlando, FL, US
www.infocommshow.org
InfoComm International is the trade association representing the professional audiovisual (AV) and information communications industries
worldwide. InfoComm 2015, the largest audiovisual show in North America, will be held in Orlando, FL, with the convention taking place from June
13–10, and the exhibits open from June 17–19. InfoComm 2015 has more than 500,000 net ft 2 of floor exhibits and special events space. More than
37,000 professionals are expected to attend the show, with more than a third of attendees coming from technology managers, specifiers, and end-
user communities.
More than 980 companies are confirmed exhibitors. In addition to the show floor and hundreds of InfoComm University sessions, Manufacturers’
Training, and specialized conferences, InfoComm has numerous networking events. The 2015 opening session at the Chapin Theater, on June 16,
will feature a panel of industry thought leaders on The Internet of Everything, moderated by Nick Bilton, Lead Technology Writer/Reporter for The
New York Times. The traditional Awards Dinner is also confirmed for the same day, recognizing honorees of Adele De Berri Pioneers of AV Award,
Distinguished Achievement Award, Educator of the Year Award, Sustainable Technology Award, Women in AV Award and Young AV Professionals Award.

July 9–11, 2015 Summer NAMM 2015

Nashville Music City Center
700 Korean Veterans Blvd., Nashville, TN, US
www.namm.org/summer/2015
With exhibits already surpassing last year’s totals, the Summer National Association of Music Merchants (NAMM) 2015 will be one of the
most vibrant shows for the music instruments and audio industry, with a strong emphasis on recording technology. The annual summer music
industry gathering is a business-friendly environment mixed with great entertainment and networking opportunities, benefiting from an amplified
presence of pro audio brands. The Summer NAMM event provides a dynamic platform for emerging and established brands to create lasting
impressions as buyers plan for the second half of the year.
Nearly 25 companies, including Marshall Amplification, Washburn Guitars, Santa Cruz Guitar Company, Zildjian, DigiTech, Seymour Duncan,
and Amati’s Fine Instruments are coming back to the show floor for the first time in several years. New in 2015, TEC Tracks offers educational
sessions focused on recording and live sound production in a dedicated area adjacent to the NAMM Idea Center.

66 | June 2015 | audioxpress.com

 Dayton Audio MCWA60 Full-range Class D Plate Amplifier
with Wi-Fi and Bluetooth 4.0 (with aptX®)
• Create or retrofit self-powered multimedia and monitor speakers
• Wi-Fi and Bluetooth 4.0 connectivity plus optical, RCA, and 3.5 mm inputs
• Fully featured wireless remote for simple, convenient operation
• 2 x 30 watts of Class D power per channel and only 4.5" wide x 6.25" high

Visit parts-express.com/axm to connect with all

of Dayton Audio’s strong, versatile plate amplifiers

Distributed By:

Tel:800-338-0531
SPA500 and SPA1000 SPA250 Watt Subwoofer SA Series Compact 725 Pleasant Valley Dr.
Subwoofer Plate Amplifiers Plate Amplifier Subwoofer Plate Amplifiers Springboro, OH 450666