Audio Critic the 16 r schematic

Issue No. 16 Spring through Fall 1991

In this issue:

Loudpeakers, still our favorite subject, are back in the
limelight, with reviews of nine different models of
widely divergent sizes and functions.

In response to the undisciplined subjectivism and lack
of scientific accountability of the high-end audio press,
our alternative audio philosophy is explicitly stated.

Our expose of the wire/cable scene continues with a
computer analysis of the effects of speaker cables.

We review a state-of-the-art consumer DAT deck and a
collection of other sophisticated electronic components.

Plus all our usual columns and features, including choice
put-downs in "Hip Boots" and a slew of CD reviews.

Retail price: $7

pdf 1

Issue No. 16 Spring through Fall 1991
Editor and Publisher Peter Aczel

Contributing Technical Editor David Rich
Cartoonist and Illustrator Tom Aczel
Business Manager Bodil Aczel

The Audio Critic® is an advisory service and technical review for consum­ers of sophisticated audio equipment. The usual delays notwithstanding, it is
scheduled to be published at approximately quarterly intervals by Critic
Publications, Inc. Any conclusion, rating, recommendation, criticism, or
caveat pubished by The Audio Critic represents the personal findings and

judgments of the Editor and the Staff, based only on the equipment avail-

able to their scrutiny and on their knowledge of the subject, and is therefore
not offered to the reader as an infallible truth nor as an irreversible opinion
applying to all extant and forthcoming samples of a particular product.
Address all editorial correspondence to The Editor, The Audio Critic, P.O.
Box 978, Quakertown, PA 18951.

Contents of this issue copyright © 1991 by Critic Publications, Inc. All rights
reserved under international and Pan-American copyright conventions. Reproduc­tion in whole or in part is prohibited without the prior written permission of the
Publisher. Paraphrasing of product reviews for advertising or other commercial pur­poses is also prohibited without prior written permission. The Audio Critic will use
all available means to prevent or prosecute any such unauthorized use of its material
or its name.

For subscription information and rates, see inside back cover.

pdf 2

49 A Brief Update on CD Players

By David A. Rich, Ph.D., Contributing Technical Editor

51 The Wire and Cable Scene: Facts, Fictions, and Frauds Part II

By Peter Aczel, Editor and Publisher

59 Hip Boots

Wading through the Mire of Misinformation in the Audio Press

59 David Zigas and Tim Smart in Business Week
59 Anent George Tice in The Absolute Sound et al.

61 Recorded Music

Mehta and the New York Philharmonic to the Max (Wilcox, That Is) and other CD reviews

By Peter Aczel, Editor and Publisher

3 Box 978: Letters to the Editor

1

Contents

11 A Loudspeaker Miscellany:

Big Boxes, Satellites, Dipoles, Subwoofers

By Peter Aczel, Editor and Publisher

11 Cambridge SoundWorks Model Eleven
12 Carver "Amazing Loudspeaker" Platinum Mark IV
14 JBL XPL160A
15 Snell Type C/IV
16 Velodyne ULD-15 Series II

21 The Minimonitor Reexamined: Four Current Examples

By David A. Rich, Ph.D., Contributing Technical Editor

22 Spica TC-50
23 Audio Concepts Sapphire II
25 Snell Type Q
26 Infinity Modulus

31 Basic Issues of Equipment Reviewing and

Critical Listening: Our Present Stance

By Peter Aczel, Editor and Publisher

35 The Electronic Browsing Section: A Collection of Totally

Unrelated Pieces of Audio and Video Equipment

By Peter Aczel, Editor and Publisher

35 Arcici Q-1 (Metal Stand for the Quad ESL-63)
35 Bryston 10B (Electronic Crossover)
36 Carver Model PT-1250 (Professional Power Amplifier)
37 Coda 01 (Preamplifier)
43 EAD "AccuLinear" (CD Player Mod)
43 Esoteric P-2 and D-2 (CD Drive Unit and Multi D/A Converter)
44 Philips LHH500 (Compact Disc Player)
45 Sony DTC-87ES (Digital Audio Tape Deck)
46 Toshiba CX3288J (32" Color TV with Surround Sound)

pdf 3

About This Issue:

Comments by the Editor/Publisher

When the history of The Audio Critic is reviewed at some point in the future, this

may turn out to have been the most important issue. Its recipients this time include not
only our current subscribers but also a much larger number of other audiophiles who are

getting this one issue as a free sample. It's a promotional idea based on my perception

that the main reason why a typical audiophile doesn't subscribe to The Audio Critic is

that he doesn't know it exists, or if he has heard of it he hasn't ever had a copy of it in his

hands. In other words, my conceit is that to see The Audio Critic is to want it. I made

sure, therefore, that such a widely circulated issue defines the editorial viewpoint of the

publication as clearly and comprehensively as possible. I and my small journal are what

you see here, warts and all.

* * *

Most regrettably, my plan to publish regularly at quarterly intervals in 1991 turned
out to be unrealizable. The last-minute unavailability of high-quality editorial help I had
been counting on was the main reason; there were also personal reasons, which at this

point are no longer in force. It's quite clear that a major operational overhaul is required

to make a quarterly schedule possible in 1992; the first steps in that direction have

already been taken. A Winter 1991-92 issue is scheduled to come out early in the winter;

the reorganization will proceed on parallel tracks. By the time the Spring 1992 issue is
due, the getting-our-act-together process should be complete and the quarterly schedule
automatic. That's the plan, and I have every reason to believe that this time it will work.

* * *

One of the consequences of all the delays in 1990 and 1991 is that the unpublished

remainder of the "Seminar 1989" transcript is a little out-of-date, at least enough so that

I'd be uncomfortable taking up a lot of pages with it. It isn't lost to posterity, however;

the words remain captured and are available for some sort of future editorial use, if and

when the occasion arises. Meanwhile, all you Stanley Lipshitz enthusiasts—yes, he has
quite a fan club out there—can enjoy the workings of that steel-trap mind once again in

the letters column starting on the opposite page. The seminar participants are still among
my favorite brains for picking, and you can expect to hear from them from time to time.

2

pdf 4

Box 978

Letters to the Editor

"When men understand what each other mean, they see, for the most part, that controversy is either

superfluous or hopeless," said Cardinal Newman in one of his famous sermons. This column attempts to

promote understanding of what audio people really mean and thereby render their controversies

superfluous or at least identify them as hopeless. Letters printed here may or may not be excerpted at the
discretion of the Editor. Ellipsis (...) indicates omission. Address all editorial correspondence to the
Editor, The Audio Critic, P.O. Box 978, Quakertown, PA 18951.

The Audio Critic:

I noticed your editorial comment
about whether damaging "Letters to the Ed­itor" get published. In July 1990, I sent you

a courtesy copy of a letter I had sent to Ste-
reophile in response to their inadequate and
misleading coverage of the AES 8th Inter­national Conference: The Sound of Audio.

Because that correspondence was com-

pletely ignored by Stereophile, I thought

you may wish to publish it in your letters
forum. [It needed far too much additional

background information to make it clear to
all comers.—Ed.]

After further reflection I wish to note
that blind testing of power amplifiers has
uncovered some rather important informa­tion about the response bias of audiophiles
(in addition to showing that properly de­signed amplifiers operated within their
power limits really do sound the same). In
every test that used the Same/Different
scoring format, listeners had a strong ten­dency to report Different when an amplifier
was compared to itself.

Approximately 35% of the time, sub-

jects in these tests heard differences when

there were none. This is an important
finding: a person with a strong interest in
audio will tend to hear differences about a
third of the time even when the devices be­ing compared are level-matched and sound

exactly the same. If things sound different
to us even when they are the same, think
about the tremendous bias toward "hearing
things" when you have a coach, such as a
salesman.

It's also interesting how the "wishful
thinking" analysis tends to persist. Martin
Colloms revisits his 1986 blind tests in the
January 1991 Stereophile. Here he recounts
how people were "shown" by statistical
analysis to have been able to distinguish
between two amplifiers. In fact, a cursory
examination of the Hi-Fi News & Record
Review article shows that while subjects
scored 63% correct when the amplifiers
were different, they also scored only 65%
correct when the amplifiers were the same.
About a third of the time they "heard
things" that could not have been there.

Colloms based his conclusions on the
correct-answer rate of the Different presen­tations alone. Had he included the Sames
and adjusted his expected score for the re­sponse bias (i.e., subject will report Differ­ent 35% of the time even when faced with
a Same), his results clearly would not sup­port the conclusion that subjects could hear
a difference.

For example, if you conducted 100
trials where amplifiers were always differ­ent, you would expect that subjects would
get 35 trials correct just because they

would tend to report differences even when
there were none. Then, if they were just
guessing, they would get approximately
32-33 of the remaining 65 right. Combined
we would expect a score of 67-68 correct.
Which is exactly what Colloms got.

How he imagines his 63% correct rate
proves his point is beyond me. I also won­der why he never answered my letter to
him raising these issues. Or why Stereo-
phile didn't publish the copy of it I sent to
them. Or why they didn't publish the letter
I sent to them about the same subject.

I also hasten to add that we should
consider very cautiously the advice given
by magazines that have been unable to veri­fy their findings under controlled conditions
and resort to voodoo statistics to imply they
have. If an editorial/review staff cannot
fairly evaluate their own tests, what would
make us think they can fairly evaluate an­other person's component?

Tom Nousaine
Cary, IL

You're not being singled out, Tom.

When Stereophile tried to make me look
like a sleazy fly-by-night in 1988, I wrote
them a letter that would have exposed their
petty ill will and irresponsibility if pub­lished. The letter was highly printable in

tone and very much to the point, but it nev-

3

pdf 5

er appeared in their pages. Selective even­handedness, right?

Your criticism of the Same/Different

method of blind testing is well-taken. The

ABX method is far better because it has no

built-in bias. The subject is asked, "Now
that you' ve familiarized yourself ad libitum
with the sound of A and the sound B, what
do you think X is? Is it A or is it B?"
There's no reason for anyone in that situa-

tion to lean toward either A or B.

As for Martin Colloms, he seems to be
selectively scientific. When it suits him, he
offers some sort of proof or technical ratio­nale for his conclusions; at other times we

just have to take his word for all kinds of

off-the-wall golden-ear assertions. I think
deep down he knows the truth, but he has
obviously pledged irreversible allegiance
to the high-end lobby, where certain truths
are totally unpalatable.

And yes, I agree, 63% is a most un­convincing score. It would be unconvincing
even as a bona fide bottom line, without the

fudged scoring a la Colloms. When A

sounds so much better than B, why can't
the golden ears score 90% or even 100%?

—Ed.

The Audio Critic:

Attention all owners of Philips DAC
960 D/A converters.

We have found that approximately
half of the Philips DAC960's we have en­countered in the field have a serious design
flaw. Specifically, the location of two criti-

cal capacitors in the de-emphasis circuitry
has been swapped by the manufacturer,
causing a significant frequency-response
aberration (a peaking of several dB in the
midtreble frequencies) in the right channel
on all CDs recorded with pre-emphasis.
This problem is easily corrected by swap­ping the locations of the 18 nF and 5.6 nF
capacitors in the current-to-voltage conver­sion stage of the right channel.

Readers with DAC960's should ask

their retailers to check their de-emphasis
circuitry with an oscilloscope and a test
CD, or telephone us at [phone].

John S. Hagelin, Ph.D.
Director of Research
Enlightened Audio Designs Corp.
Fairfield, IA

A number of weeks after having been
alerted to the above, Philips Consumer
Electronics responded as follows:

The Audio Critic:

After contacting both our factory and
our service center concerning the alleged

4

"design flaw" of our DAC960, as reported
by Mr. Hagelin, I am pleased to report that,
based upon inspections of our stock, no
mistakes were found. We will gladly repair
any units found in the field that were inad­vertently produced incorrectly, though we
do not believe this is an issue.

Best regards,
Mike Piehl
Philips Audio Marketing Manager

The Audio Critic:

Dear Dr. Rich,
We received your letter inviting a re-

sponse to the article "The Present State of
CD Player Technology: Who Is Doing It
Right?"

We expect to have a few "arrows di­rected our way" because of the fact that our
basic assumptions and methods are radical­ly different from the norm in high end digi­tal decoding. Our research team and indus­try colleagues recommend that I give a
brief explanation of "where we are coming
from" and encourage free and open debate
on the critical issues of digital decoding.
Please understand that in responding to a
few "arrows," we are only attempting to
give an explanation for the trade-offs we
have made in the design of our products—
we are not attempting to defend our posi­tions nor do we expect to "win you over,"
so to speak, since your positions appear to
be quite polarized in many respects.

One major point of interest, as you
well know, is the importance of a monoton­ic decoding algorithm. You state that our
algorithm has "not been optimized for max­imum passband flatness." (It should be not­ed that we have no passband ripple, which
is sometimes confused with maximum
flatness.) Many people argue in favor of a
flat frequency response (no slight droop at
20 kHz). Fine with us, but only if and when
that can be done with an algorithm that is
also monotonic. (You stated the trade-off

but failed to give your readers any idea as
to why this trade-off was made. Even if
you and Dr. Lipshitz don't personally "be­lieve in" monotonicity, it would be only
fair to explain to your readership that we
do.)

Axiom #1: In the meanwhile, it is

more important to great sound that the de­coding algorithm be monotonic than that it
have a perfectly flat frequency response.

It is so easy to kick up the response at
the upper end, in order to achieve good
specs. But it is not presently possible to do
this and also remain monotonic. If the re­sponse is not monotonic, then you have
passband ripple, echoes and, therefore,

TDE (Time Displacement Error). Passband
ripple would probably need to be 0.000001
dB or less not to cause time-based distor­tion within the digital filter, where math is
often done at 36-bit resolution and then
truncated.

Axiom #2: TDE is the most critical
parameter separating great analog perfor­mance from the performance achieved
from conventional digital decoding. (Again,
even if you and Dr. Lipshitz don't personal­ly see the importance of TDE or agree that
conventional digital has a serious amount
of the wrong kinds of it, a brief explanation
of our position would have been helpful to
your readership.)

Your readers may not know that some
of our researchers have been studying the
effects of TDE on audio for up to 30 years.
Dr. Robert Bradford, our Chief Technical
Officer, literally wrote the book on TDE as
it relates to professional recording. We are
enclosing a couple of the technical deriva­tions he did for 3M-Mincom, to illustrate
the fact that our team has great technical
depth on this subject.

Last week, we had two Wadia VPs
and a designer at my listening room audi­tioning a new phono cartridge we had just
installed and adjusted in my system. Again,
we reconfirmed the fact that we like the
sound of good analog. It is involving and

pleasing to listen to. We have found in re­peated tests, over several years, that the
only way digital decoding can compete
with good analog in performance is by use
of algorithms and techniques that reduce
the time distortion to "about zero." TDE is
the central issue in high-end digital audio.

It should be noted that we, as an engi-

neering community, knew about this in
many different ways, for many years. I first
ran into TDE in 1964. At that time, M-A-K
Inc. had the idea of using a Cray-1 (under
development in central Wisconsin) and
packet switching methods to produce a tele­phone digital switching system. Packet
switching was abandoned, however, as a
suitable technique, and later replaced with
a time-accurate transmission topology by
Bell Labs, due to the subjective sonic irrita­tion of TDE. It was in the Bell Labs Blue
Book where I first saw the warning that
"amplitude ripple (i.e., the lack of monoto­nicity) causes TDE." In spite of all this, the
mass-market digital people are choosing to
ignore time-based issues in digital decod­ing. This is so they can continue to use in­expensive and simple sin x/x decoding
methods. Wadia believes these "ripple de­coders" are fine for the mass market, but
we are alarmed to see such techniques mas-

pdf 6

queraded as high-end techniques. We are
even more alarmed that the high-end press
is allowing them to get away with it.

On the subject of "ripple decoders" we
again believe your readers would have ben­efited from hearing both sides of the story.
You state that the lack of precursor ripple
on our transient response is due to the
shape of our "filter," which leads to out-of­band image energy. What you didn't say
was that the converse is also true. The pres­ence of precursor ripple energy on competi­tive products' transient response is the re­sult of their brick-wall filter shape. That
ripple is not there for any good reason; it is
a result of a filter shape. We think people
should know that a brick-wall filter has
about the same amount of ripple energy
(you can see it clearly on a scope), during
transients, right at band edge, as we have
image energy above band edge. But the
press only mentions image energy because
it can be seen on a spectrum analyzer,
whereas ripple energy averages out over
time and is not seen on a spectrum analyz­er. Even though it may average out over
time, the instantaneous time-distortion
damage is still very real. (We understand
that you and Dr. Lipshitz disagree with us
over the importance of precursor ripple vs.
the importance of image energy. Even so,
you could have stated our position so your
readers understand why we made this
trade-off. The reader might think the only
kind of band-edge or out-of-band energy is
image energy.)

What about our low-level linearity?

It is much less than one CD LSB and
is, therefore, very good. It is off about 2 dB
at -90 dB (5 microvolts of error, ref. 0
dBV). This has no sonic significance. (You
made it sound alarmingly high! Why?)

Since you brought it up, what did we
really say about the sampling theorem and
the Fourier series?

First, we believe it is important for
people to know that the sampling theorem
clearly implies that once a musical signal is
sampled (all chopped up into numbers), it
is impossible to get it back "perfectly" ac­curate again. One reason is that the sam-

pling intervals at the A/D and D/A must be

"perfectly" the same, which is impossible
(e.g., the jitter problem you and others have
addressed). Another reason is that we do

not have access to plus and minus infinity.
The focus of our research is to get as close

to perfection as possible where it really

matters.

Second, we constantly find it neces­sary to remind people that Fourier never
said "the world" was made up of sine

waves (implied, perhaps, by the Fourier se­ries). We simply remind them that this

worldview is only an "approximation," as

is clearly seen by the fact that "any theory
that relies on an infinite series is, by defini­tion, an approximation." There are many

infinite series that can be used to model
"the world." We have publicly stated that
the Fourier series is not always the opti­mum approximation (especially in the case
of inharmonic and transient musical wave­forms), and we often prefer to use other
approximations that are judged to be more
appropriate for the design task at hand. For
example, we often find it convenient to as­sume that "the world" is made up of an
infinite number of impulses spaced
infinitely close together. The Fourier series
is only one of many mathematical tools, to
be used as appropriate. It is not sacred!

(Dr. Rich, we feel our view is very
reasonable. Your comment about this in
your article appeared as though you were
really "out to get us"! Why? Do you have a
favorite infinite series that you are promot­ing today? This reminds us of some of the
arguments we have had with Dr. Lipshitz.
We just don't get it!)

It is interesting that you use the refer­ence [Papoulis 1984] to refute what we say,
while Abel Graham ("What's Critical in
Digital," enclosed) uses Papoulis to help
prove our point on TDE. Bradford suggests
that this same Papoulis ("The Fourier Inte­gral and Its Applications," 1962) was prob­ably the first academic to warn us in high­end digital audio that passband ripple (the
lack of monotonicity) leads to time distor­tion in transient response. I guess I had bet­ter go back and read the references again,
to see whose side he is really on. (Just kid­ding.)

(Please understand that the purpose of
this letter is to explain "where we are com­ing from." Taken out of context, many of
the above statements would appear to be
defensive or blatantly self-righteous. Please
don't quote us out of context. As we see it,
we are all struggling with the same issues
and fighting the same technical battles. We
simply approach certain problems from dif­ferent vector angles, depending upon our
experience and background.

So, keep up the good work, and keep
the "debate" alive. As they said back in the

'60s: "Let a hundred flowers blend, let a

hundred schools of thought contend."

Yours truly,
Don Moses
[CEO]
Wadia Digital Corporation
River Falls, WI

Dr. Rich replies to Don Moses:

I am disappointed that you have cho­sen to give your standard "manufacturer's
comment" reply to my article and have not
addressed the points I made in the article.
You continue to state that the sampling the-

orem is valid only for deterministic sine

wave signals. As I discussed in my article,
the sampling theorem is equally valid for
stochastic signals, such as music. You con-

tinue to ignore the presence of a brick-wall

antialias filter at the input of the analog-to-

digital converter.

You state that your time-domain inter-

polation algorithm is superior to conven­tional techniques, although you have not

supplied any data to justify your claims.
Since the interpolators are operating in the

digital domain, it should require a trivial

amount of work to show that your method

yields a smaller minimum mean square er­ror (or a smaller error by any other error
criterion) than the frequency-domain meth­ods. The original objective of the work by
Robert W. Moses (who is apparently not
employed by Wadia), as stated in the MON-
TECH paper, was to find the optimal filter
coefficients that would minimize the error
in the interpolated data. According to the
MONTECH paper, a time-domain algo­rithm was chosen because the optimum
coefficients that would minimize interpola­tion error could be more easily calculated

in the time domain. The paper makes no

mention of TDE. Your axioms are nowhere
to be found in the MONTECH paper. I be­lieve that, given a sufficient number of
taps, the time-domain optimization would
yield coefficients very similar to a filter de­signed in the frequency domain. It cannot
be guaranteed that the resulting optimized
filter would have the monotonic property
required by your first axiom, but you have
sent no analytical explanation to justify the
axiom.

You enclosed a number of papers by
Dr. Robert S. Bradford on TDE. These pa­pers analyze the effects of flutter and asso­ciated time-base errors on the performance
characteristics of an analog tape recorder.
This has no relevance to the interpolation

and smoothing of digital signals. Dr. Brad­ford's work can be extended to include the
effect of time-base jitter on a sampled sig­nal. In this extended form, Dr. Bradford's
work would clearly indicate the need for a

small peak-jitter error in the recovered clock

signal. As can be seen from my original ar­ticle, I would not dispute this conclusion.
Perhaps you have not fully understood that
Dr. Bradford's work relates to the clock jit­ter problem and not the reconstruction of

5

pdf 7

the digital signal. In the preprint of the
Abel Graham newsletter (it is unclear what,
if any, part of this preprint was published,
and no one I know in the electronics indus­try has ever heard of this newsletter), TDE
is discussed in the context of the interpola­tion and smoothing of digital signals. Mr.
Graham has apparently also not fully un­derstood Dr. Bradford's work.

The sole purpose of a reconstruction
filter is to remove the image energy from
the output of the DAC. If you believe that
the magnitude of the image energy at the
output of your decoder is unimportant, then
the entire reconstruction filter can be elimi-

nated. With the reconstruction filter elimi­nated, the shape of the digital impulse
would show no ringing.

With regard to the droop at 20 kHz, I

believe this is quite audible and will cause
the Wadia player to sound less bright than
competitive players. The droop could be
corrected in the analog section of the player
if you did not want to modify the digital
filter. I note with interest that you make no

offer in your letter to allow The Audio Crit-
ic to evaluate a Wadia product. If a unit

were made available, I would perform a
simple listening test. The test would be as
follows: Encode an analog source with a
digital recorder and compare the sound of
the reconstructed output with the original

source. If the digital recorder is of good

quality, no difference will be heard in an
ABX test. Now replace the digital record­er's DAC section with the Wadia decoder. I
believe that the Wadia will be clearly audi­ble in the ABX test. This test would con­clusively show that the Wadia decoder is
changing the sound of the original source.

With regard to lowlevel linearity, I

indicated in my article that gain linearity
provides only a limited amount of informa­tion on the DACs performance. Harmonic
distortion measurements are much more
important. A Wadia X32 was found to have
30% harmonic distortion at 90 dB by Ste
reophile (Aug. '90, Vol. 13, No. 8, p. 125).
Competitive stateoftheart products are
now using DACs with almost unmeasurable
harmonic distortion at 90 dB (PS Audio,
Theta, Meridian, and Harman/Kardon, for
example). I am amazed that you find it ac­ceptable to produce a $7995 decoder box
which has poorer lowlevel harmonic dis­tortion performance than a $200 CD player
with MASH DACs. The effects of low
level linearity errors are the only significant
measurable differences between modern
CD players. I am surprised that you find a
passband ripple of 115 nV unacceptable
(the increase in magnitude of a 1 V rms sig

6

nal subjected to a 0.000001 dB level in­crease) yet find a 5 µV gainlinearity error
(relative to 1 V rms) reasonable. Since Wa­dia uses highquality DACs, it is quite pos­sible that the lowlevel linearity problem is
the result of a software "bug." That would
be good news for Wadia owners because,
once the bug is identified, the problem
could be fixed by changing an EPROM.

Finally, I want to point out that Dr.
Stanley Lipshitz reviewed my manuscript;
he was not a coauthor. Any similar com­ments made to you by Dr. Lipshitz regard­ing your product's design are independent
of my analysis.

David Rich
Contributing Tech. Ed.

Dr. Lipshitz replies to the Editor:

I would like to respond to Don Moses'
letter for three reasons: (a) Although I am

not the author of the article on which he is
commenting, he repeatedly addresses his
remarks to both David Rich and myself
("you and Dr. Lipshitz"); (b) I agree with
most of the statements made by David Rich

in his article, and in particular with those

concerning Wadia's decoding algorithm, to
which Don Moses takes exception; and (c)
I have on numerous occasions expressed to
Don Moses my belief that his company is
fundamentally misguided in its digital filter
design.

The first point to note is that, in the an-

alog reconstruction process, one is attempt­ing to recover as accurately as possible the
original analog signal whose samples have
been recorded. Of the infinity of such ana­log signals (yes, there are indeed infinitely
many signals which have the same sam­ples; they are all aliased versions of one an­other), there is only one which is bandlimit­ed to the Nyquist frequency (one half of the
sampling frequency). This unique analog
signal is the one which we should be trying
to reconstruct. It is the bandlimited signal
whose samples were taken in the original
analogtodigital conversion process. (The
input antialiasing filter did the required ini­tial bandlimiting.) These statements are the
essence of the sampling theorem. The earli­est proof of the sampling theorem of which
I am aware was given by E. T. Whittaker
[1] in 1915. Whittaker presents a very gen­eral and profound result which includes the
proofs of the statements made above. The
uniquely represented analog signal is what
he calls the "cardinal function." Now, the
essential point is that the cardinal function
is obtained from its samples by a sin x/x re­construction process, this being the time
domain equivalent of a brickwall bandlim-

iting filter set at the Nyquist frequency in
the frequency domain. (The Fourier trans­form of a perfect brickwall filter is a sin x/x
function.) The process of removing the
"images" of the Nyquist band by means of
this brickwall filter results in the recon­struction of the originally sampled analog
signal. This is a mathematical theorem.

I must thus reject Wadia's claim that
there is something improper or deficient or
inappropriate in trying to approximate as
closely as feasible a true sin x/x reconstruc­tion. This ideal reconstruction filter (which
can only be approximated) must pass with­out change all frequency components up to
the Nyquist frequency (i.e., it must have a
flat passband with linear phase response up
to the Nyquist frequency) and completely

remove all frequency components (the "im­ages") above this frequency. It must thus
approximate to a brickwall filter, and the
extent to which it fails to do this is a mea­sure of the error it makes in the reconstruc­tion. Note, by the way, that it does not mat­ter whether the brickwall filter is all
analog (as in early digital audio systems),
or partly digital and partly analog (as in
current systems). It must be there.

This brings me to my second point.
Given the above, why does Wadia use a re­construction filter which significantly atten­uates the high audio frequencies (by 3 dB)
and passes a goodly chunk of the outof
band images? Moreover, why do they
maintain that their reconstruction is more
accurate than a sin x/x reconstruction?
(More accurate to what?) I believe that it is
a misguided approach, based on an approx­imation to the wrong criterion. A cynic
might be inclined to speculate that the tre­ble cut masquerading as greater accuracy is
the audible reason why some people might
"prefer" this less accurate (to the original
samples) sound. But inaccurate it is. It does
not come close in any sense to approximat­ing the ideal brickwall filter discussed
above. Most digital audio reconstruction
filters are much closer. So what lies behind
Wadia's filter design? It is an attempt to
make the filter's impulse response more
compact in time than the ideal sin x/x
filter's oscillatory timedomain behavior.
This attempt seems to be based on the be­lief that there is something inherently
wrong with the latter, but as the sampling
theorem shows, this is not the case. The
pre and postringing of the ideal brickwall
filter does not in any way introduce precur­sors or postcursors (?) which were not al­ready present in the original bandlimited
signal which is being reconstructed from its
samples. To believe otherwise is a serious

pdf 8

misunderstanding of the mathematics in­volved, and hence of the true outcome. This
may seem counterintuitive, but it is correct.
For example, if the input analog signal was
bandlimited by a causal brick-wall filter ap­proximation (e.g., a minimum-phase analog
antialiasing filter), which thus had no pre­cursors in its impulse response, an ideal
sin x/x reconstruction will not introduce any
precursors .

It seems that Wadia believes that the

ringy nature of a sin x/x impulse response is

inherently undesirable (whereas, as I argue

above, it is actually correct), and so sets out
to design a filter with less pre- and post­ringing, which is what they have done. But
you cannot have it both ways. To the extent
that your filter departs from a sin x/x im­pulse response it is reconstructing a
modified (read "wrong") version of the sig­nal. In their pursuit of a filter time-domain
response closer to the mistaken goal of a
perfect nonbandlimited impulse with no

overshoot or ringing, they have attenuated
the top half of the audio band and also al-

lowed substantial ultrasonic garbage out

(which is nonlinearly related to the original

analog signal). If Don Moses really be-

lieves that a single-sample-high impulse re-

sponse is ideal, he can very easily achieve

it. Just omit the reconstruction filter entire-

ly, and allow the baseband and all images
out unattenuated! Why does Wadia not do

this? Because then, of course, you don't get

an analog-looking signal back—you get
back the sampled waveform with all its dis­continuities, a far cry from the original.

To summarize, I maintain that Wa­dia's approach to digital-to-analog conver­sion is inherently flawed because of what
appears to be a misunderstanding of the
sampling theorem itself. Don Moses' desire
for a monotonic frequency response would
seem to be simply a reflection of an unnec­essary constraint, which forces his system
into the errors that it makes as a result of
the Lagrangian interpolation used. In no
way do I accept Don Moses' two "axioms."
Maybe Wadia ought to reassess them.
(Note that an axiom is not a provable re­sult, but an assumption from which results
can be deduced.) Finally, the sampling the­orem relies, not on the Fourier series as

claimed by Don Moses, but rather on the

Fourier integral, which does imply "an
infinite number of impulses spaced
infinitely close together." There is no con­tradiction inherent in the use of the sam-

pling theorem.

Yours sincerely,
Stanley P. Lipshitz
Audio Research Group

Departments of Applied Mathematics

and Physics
University of Waterloo
Waterloo, Ontario
Canada

Reference:

[1] E. T. Whittaker, "On the Functions
which are represented by the Expansions of
the Interpolation-Theory," Proc. Roy. Soc.

Edinburgh, vol. 35, pp. 181-194 (1914-

1915).

As Editor, I want to make absolutely
sure that the reader understands the essen­tial thrust of the professorially restrained
commentary by the two academics above.
To put it less politely but more simply than
they do, Wadia Digital is designing and
selling D/A conversion equipment based on
incorrect mathematics. The Moses versus

Rich/Lipshitz debate is not about some kind

of legitimate diversity of informed opinion
but about mathematically provable fact.
(By the way, that bit about "a hundred

flowers"—that's Mao in the '50s, not

"they" in the '60s.) Assuming that Moses is

presenting the rationale of his technical

team accurately—and there's always the

possibility that he isn't—Rich and Lipshitz

are clearly right, Moses is clearly wrong,
and Wadia DIA conversion is clearly

faulty. It's as simple as that—as long as

Moses puts it as simply as he does—and no
authoritative scientific opinion exists to the
contrary.

—Ed.

The Audio Critic:

David [Rich],

...[Regarding] your article: The main
point of contention I have is with your
comment that "the jitter level of the

YM3623 is sufficiently low...." It is hardly

sufficient. The jitter problems I alluded to
earlier were indeed the result of using the
YM3623 in the manufacturer's recom­mended circuit. There are tricks one can
play, however, to adjust the circuit around

the YM3623 to improve its performance,
but the part, by itself, is basically junk. Un­fortunately, it is also the only low-cost
commercially available S/PDIF interface
chip out there, so I guess we have to live

with it until Yamaha, Crystal, and Philips

all get their PLLs to work. (The new Yama­ha receiver is over a year behind schedule
because of this problem, and the Crystal
part is overdue as well, I suspect for the

same reason.)

Secondly, your statement that a
"brick-wall" filter has a sin x/x response is
patently false. The impulse response of any

system is, of course, the frequency re­sponse of the system; for the filter, it would
be the filter's response characteristic. The
impulse response of the sampling system is
sin x/x, in consequence of the finite sam­pling time. So let's put the blame where it
is due, and leave my poor analog filters
alone!

Power supply bypassing, rather than
separate regulation, is all that's necessary
for good performance on our DACs. The
problem we've found most people have is
that they use poor bypassing and layout
techniques—and often end up using regula­tors as a crutch to solving their problems.
This is an extra expense that could have
been saved if they took the time to lay out
their circuit properly in the first place.

Other than these minor points, your ar­ticle was comprehensive as well as well
written. I'm even thinking about subscrib­ing to The Audio Critic, after having seen
this issue. It seems to be the only "audio­phile" magazine I've seen that takes a prag-

matic approach to audio....

With best regards,

Rick Downs
New Product Development Engineer
Audio Products
Burr-Brown Corporation
Tucson, AZ

Dr. Rich replies:

The brick-wall filter I referred to was
the ideal brick-wall reconstruction filter re­quired by the sampling theorem. An ideal
brick-wall filter has constant magnitude in
the passband below the cutoff frequency f

c

and total rejection above the cutoff frequen­cy fc. A lowpass filter of this form (which
is not realizable) will have a sin x/x time­domain impulse response. The sin x/x time­domain impulse response in the sampling
theorem is a result of this ideal reconstruc­tion filter. As Mr. Downs indicates, a real­izable brick-wall filter will not have a sin x/x
response. The sin x/x response is closely
approximated by the digital filters used in
modern CD players.

The sin x/x frequency response which
Mr. Downs refers to is a property of a real
DAC that I did not discuss in my original
article. In the ideal sampling theorem, the

sampled signals are assumed to have the
form of an impulse function. In practice,
each sample at the output of the DAC has a

finite width (Mr. Downs refers to this as the
finite sampling time) and is approximately
rectangular in shape. This characteristic of
a real DAC can be modeled as an ideal
DAC followed by a fictitious filter. The fre­quency response of this filter will have a

7

pdf 9

sin x/x response. This additional filter re-

sponse results in a small high-frequency
rolloff. This drop is compensated by the
digital or analog filter in a CD player.

David Rich
Contributing Tech. Ed.

The Audio Critic:

...You saved me a lot of money, as I

was about to purchase a Meridian 208 CD

player/preamp.

I learned a great deal from Dr. Rich's
essay but was disappointed that the CD re­views said that it makes no difference what
is inside the players—they all sound alike.
Did some players not show any difference
in soundstage or image focus? What about

depth??? Yes, you saved me a lot of mon­ey. But now I don't know what the hell to
buy. Thanks.

Ralph Riutti
Moorpark, CA

Let's be precise. I never said it makes
no difference what's inside the players; on
the contrary, I discussed at some length the
measurable differences in electronic per-

formance, as well as the differences in

construction quality and ergonomics. What
I did say was that, within the group of 13
units reviewed in that particular issue (No.

15), I and my associates found no audible
differences in the course of a somewhat
limited number of ABX comparison tests.
On a previous occasion (see Issue No. 12,

p. 36), we heard, and I reported, a tiny dif-

ference in one instance.

As for soundstage, image, depth, etc.,

you must understand—as the audio pundits

whose golden ears are attached to muddled
heads do not—that those "structural" qual­ities of reproduced sound are determined

by the recording site, the microphones, the
microphoning and mixing techniques used,
the signal processing added (if any), and

the radiation characteristics of the play­back loudspeakers—not by the design of
the playback electronics. You can safely as-

sume that a reviewer who waxes eloquent
over the soundstaging or depth of an elec­tronic circuit has no serious credentials as
a technical expert. Even the easily measur­able difference between 15-bit and (almost)
16-bit resolution in digital audio appears
to be audible only on special test signals
and not on music.

—Ed.

The Audio Critic:

I would like to add a few personal

comments to the discussion between your­self and William J. Roberts [Issue No. 15,

8

p. 7] on the subject of constant-directivity
speakers.

There is no connection between how
something is recorded and how it is repro­duced. I have never seen a report of some­one changing from a pair of bipolar speak-

ers, used to listen to music recorded with
ribbon microphones, to a pair of omnidirec­tional speakers to listen to music recorded

with omnidirectional microphones, or

changing to a pair of cardioid-pattern

speakers (there are such things) to listen to

music recorded with cardioid microphones.

Likewise, there is a similar lack of re-

ports of stacking speakers on top of each
other to listen to music recorded with coin­cident microphones, or moving them apart
to listen to music recorded with spaced
microphones.

There is no connection between how

music is monitored during recording and
how it should be reproduced.

In the production process, there are

such techniques as LEDE™ and RFZ™ in
use. These are two of the several methods
used in trying to listen only to direct arriv­als when analyzing sound. One attempts to
absorb the room reflections in foam materi­al, and the other attempts to steer the direct
and reflected sound so that reflected sound
arrives too late to be perceived. To these
may be added the use of the famed 604
monitor speakers and the use of head­phones, both of which provide a preponder­ance of direct sound and little or no

reflected sound.

The reason for this is that reflected

sound results in what may be called spa­ciousness, low interaural cross-correlation,
or diffusivity, which are all related con­cepts and which correlate with listener pref­erences.

In my opinion, it is difficult for most
people to find fault with something which
gives them pleasure, which is generally a
useful trait but does interfere with the busi­ness of monitoring recorded sound, using a
reproduction system with characteristics
which listeners find pleasant.

Mr. Roberts mentioned the research
results of Floyd Toole. One of Dr. Toole's
findings was that the preference of his lis­teners correlated positively with measured
increases in beamwidth and with measured
constancy of beamwidth with frequency.
He did not claim to test anything that was
called "constant directivity," but neither did
he mention any complaints of "brightness"
going along with increases in beamwidth
and beamwidth constancy.

Yes, a measurement microphone will

give a higher reading at higher frequencies

when measuring the output of speakers
which do not get too beamy at high fre­quencies, since it picks up both direct and
reverberant fields. Simplistically speaking,
human hearing responds to direct sound for
amplitude information and to reflected

sound for spatial information. So, an in­creased high-frequency reverberant field
should not make an amplitude difference.

There have been a few consumer loud­speakers which were actually constant­directivity speakers, although not identified
as such. They were favorably reviewed but
did not seem to remain in production for
long. An example is the Genesis 44, which
seems to me to have been quickly replaced
with a more profitable "improved" model,
to which one could add examples such as
the AR MGC-1 and the original dbx
Soundfield, which also had a cardioid-like
pattern. "Brightness" was not a complaint.

Speakers with constancy of dispersion
angle do have a problem, though, related to
perceived localization.

Much classical music is recorded with
spaced microphones. This results in a re­cording which sounds just fine when played
back on beamy speakers. The localization
is adequate to identify a source location for
the string section, and there is plenty of
what seems to be " hall sound."

But, when played on nonbeamy speak-

ers, the sound field takes on an unbeliev-

able shape, or rather lack of shape. For in­stance, a violin solo is "right over there,"
and also other places. There is no "edge" to
the sonic image, so that it sort of "blurs
away" towards the sides. In other words,

the phase differences that result in pleasant­ly low interaural cross-correlation from
beamy speakers result in a loss of localiza­tion information from nonbeamy speakers.
This is simply unacceptable to some people.

There is a solution: learn to like music
recorded with "original instruments,"
which were less loud, used in smaller spac­es to produce the same volume as later in­struments, and generally recorded from a
single point, for practical reasons. Then,
speakers that are nonbeamy are not bother­some. In fact, I rather like the ones I have.

Regards,
James P. DeClercq
Roseville, MI

You oversimplify. Yes, the playback
geometry in standard practice is totally un­related to the recording geometry—but no,
that's not necessarily a desirable situation,
nor is it invariably the case. For example,
the original Edison phonograph was a sys­tem of sound reproduction in which the

pdf 10

recording geometry and the playback
geometry were perfect mirror images of
each other—and that was the best part of
an otherwise highly limited system. The

same kind of symmetry exists today in bin­aural recording and playback: the head­phones replace the microphones in exactly

the same position, without any change in

geometry. Another symmetrical technique

is to take a small group of, say, 8 perform-

ers and close-mike each of them with 8 sep­arate microphones feeding 8 separate

channels and tracks, then play the 8-track

tape back through 8 speakers deployed in

the same relative positions as the original
performers. It isn't practical and it's rarely

done, but it can be very lifelike indeed.

As for monitoring during recording,
the frequency response of the monitor
speakers obviously affects the decision of
the producer as regards the correct fre­quency balance. If the monitor speakers
have a rolled-off top end, an inherently
overbright recording will sound just fine in
the control room but not in the home
through flatter speakers. That much of a

"connection" between monitoring and

playback at home is self-evident; as for the

relative brightness of constant-directivity
speakers, I'll admit that in a large, well-

padded room the issue may be moot, but in

a small or medium-sized room with hard
surfaces the increase in reflected high-

frequency energy will not be sufficiently

separated from the direct sound to avoid
the impression of increased brightness—I

have experienced this myself

I'm inclined to agree with you, on the

other hand, on the subject of spaced micro-

phones and the trade-offs they entail. I can

live with those trade-offs, however, espe­cially since single-point microphoning has
its own characteristic shortcomings.

—Ed.

The Audio Critic:

...I was delighted with the contents of
Issue No. 15. The article by David Rich on
CD player technology is, in my opinion,
the finest article on the design of a CD
player I have read and one of the best on
audio technology I have ever seen. One of
the major virtues of this article is that it is
not condescending nor oversimplified nor
obtusely technical. I read it as I would a
good mystery—on the edge of my seat.
What I have learned from it makes me feel
one up on the trash that appears elsewhere
in the audio press and the pseudosophistica-

tion of some high-end sales people....

I have been following the double­blind test desert storm for some time. The
people who oppose it or question its validi­ty remind me of the pharmaceutical-firm
vice presidents who fought this same ap­proach for testing new drugs—they are ei­ther dumb and/or know that they have a lot
to lose from the objectivity that is forced on
them by double-blind studies. The basic
emotion common to audio gurus and phar­maceutical manufacturers in this context is
greed. Without double-blind studies a lot

more dangerous or useless drugs would be
in circulation. Too bad that there is no FDA
for audio. I write that tongue in cheek, as it
takes ten years and $10 to 100 million to
market a new drug. Perhaps an Audio Critic
would suffice for us lovers of music.

Cordially yours,
Steven E. Mayer, Ph.D.
Nashville, TN

/ really don't believe that the "audio

gurus" you refer to are motivated primarily
by greed. There are better arenas for greed
than high-end audio. (Try a used-car lot or
a massage parlor.) No, you're talking
about self-indulgent, posturing little people
looking for groupie approval and protect­ing the belief system of the cult. They're
more worried about their ego than about
their money—although vested interest can't
be entirely ignored—and they aren't big
enough to admit they were wrong even
when the facts are incontrovertibly demon­strated to them. On the contrary, the more
the scientific audio community snickers at
their voodoo, the more they try to prosely­tize those who know even less than they do.
You say they're greedy and manipulative; I
say they're untutored, fuzzy-minded, inse­cure, and unaccountable.

About David Rich's article, I agree
with you 100%. He is a great addition to
our staff. And that's just the beginning; oth­er highly accredited people will soon be
coming on board.

—Ed.

Who Sued Whom and Why: Stereophile and Carver Corporation in Court

I keep getting all sorts of inquiries
about last year's mysterious lawsuit between
Stereophile and the Carver Corporation.
Most of the inquirers are under the impres­sion that Carver sued Stereophile. Not so.
Maybe Stereophile would like the audio
community to believe that, but that's not
what happened. No audio manufacturer in
his right mind (except, of course, Bose)
would want to be the plaintiff against the
free press in a suit about a bad review. The
actual fact is that Stereophile sued Carver.

Why did they sue? Basically because
they didn't understand what they were get­ting themselves into. Carver had run a 12­page ad in the May/June 1990 issue of The
Absolute Sound, in which six Carver am­plifier reviews were reprinted. One of them
was Robert Harley's hatchet job on the
Carver "Silver Seven-t" in the January

1990 Stereophile; the five others were high-

ly favorable reviews, including one of mine.
The ad made Stereophile look kind of stu­pid—like the schoolyard bully. Thereupon
Stereophile sued Carver Corporation for
copyright infringement, claiming the latter
had no right to reprint the review without
permission. Carver responded the standard
way, by filing a countersuit.

Stereophile's suit was essentially friv­olous and fell on its face in court—the text
of an attack is not protected by copyright
when reprinted in the context of a defense
against it. Carver's countersuit, on the other
hand, had some legal substance and threat­ened to bankrupt Stereophile if pursued to
the finish. The claim was that a systematic
pattern of maliciously discriminatory
"Carver bashing" and recklessly irresponsi­ble/incompetent equipment reviewing had
caused multimillion-dollar losses to Carver
Corporation. It could have turned into a

First Amendment battle, but the court or­dered the parties to go into arbitration first.

The outcome was a rather astonishing
settlement, "with prejudice." For three
years—1991, 1992, and 1993—Stereophile

is forbidden to put the word "Carver" in
print, just about regardless of context, with
minor legalistic exceptions. They basically
have to pretend that Carver doesn't exist.
Carver, in turn, is forbidden during those
three years to discuss publicly the alleged
deficiencies of Stereophile's equipment
testing or to disseminate reprints of the dis­puted Stereophile reviews, again with some

additional minor legalisms.

The gist, as I interpret it: Stereophile

signed away its First Amendment rights to
prevent a possible disaster in court and in

exchange received what was most impor­tant to it—silence on the touchy subject of

its competence. A class act, eh? —Ed.

9

pdf 11

A Loudspeaker Miscellany:

Big Boxes, Satellites, Dipoles,

Subwoofers

By Peter Aczel

Editor and Publisher

While there are no ultrahigh-end models in this group (the most you

can spend here is two thousand and change), the best of these units

raise serious doubts about the need for insanely expensive speakers.

Those who are familiar with my loudspeaker reviews
know that I like to make some deep philosophical remarks
and broad generalizations before I get down to the nuts and
bolts of specific models. Now, I can't keep repeating myself

at the beginning of each new loudspeaker survey (like this

one) just to communicate to new arrivals where I'm coming
from philosophically, so I must refer first-time readers to
earlier issues, especially Nos. 10, 11, and 14. I do want to
bring up here, however, a point I haven't made perfectly
clear (if you'll pardon the tainted expression) before.

Where are the curves?

When it comes to speaker systems, I don't particularly
like to show frequency response curves and other graphic
displays of performance. A frequency response curve is fine
and dandy for showing, say, the de-emphasis error in a CD
player or the characteristics of a parametric equalizer, but it
can be quite misleading in the case of a loudspeaker system.

A difference of a few inches in microphone placement can

make a tremendous difference in the measured response

curve—very flat and smooth this way, quite jagged that
way—and the audiophile looking at the curve in a magazine
will jump to erroneous conclusions either way. Rigid mea­surement protocols—such as aiming the microphone at the
geometric center of the speaker system, or at the tweeter, or
at the woofer, from a distance of one meter, three meters,

etc.—will result in superficial and inconclusive data. Each

speaker system tends to be a law unto itself and must be
measured with a certain flexibility of technique that comes

from experience. Formularized measurement with pat

graphic output as its goal is poor audio journalism, at least

in my opinion. (When loudspeakers become as predictable

as amplifiers, I'll change that opinion.)

My method is to use the B&K microphone more or

less as a doctor would use his stethoscope, poking and prob­ing every which way, near and far, at the "sweet spot" and

at the not-so-sweet spots, using all sorts of test signals and
monitoring everything on the spectrum analyzer and/or the
oscilloscope. Pretty soon I have a very good idea of just
how smooth the response is, whether there are trouble spots
(ringing, lobes, phase reversals, etc.), how deep the bass
goes, whether the output from the various drivers coalesces
into a semblance of coherence and at what point, and so
forth. No, it's not as perfect a technique as I would like—
and, yes, I do take fixed on-axis and off-axis measurements
at set distances, but I don't entirely trust them. A large and
systematic family of curves taken in an anechoic chamber,
which I don't have, would probably be preferable but still
subject to audiophile misinterpretation if published; the gated
pseudoanechoic measurements favored by some reviewers
also have serious limitations. I contend that my eclectic
method arrives at the qualitative truth—and isn't that the
reviewer's truth?—without fail and with a high degree of
objectivity, even if it leaves something to be desired quanti­tatively from the engineering researcher's point of view.

Oh, yes, in all fairness, there is a routine, formularized
measurement which is very accurate, namely the Don Keele
method of extreme-nearfield bass response measurement. It
tracks the anechoic curve beautifully up to 100 Hz or so,
and I bow toward Elkhart, Indiana, every time I avail my­self of Don's great little shortcut. Wouldn't it be nice if it
worked equally well at higher frequencies?

Cambridge SoundWorks
Model Eleven

Cambridge SoundWorks, Inc., 154 California Street, Newton , MA

02158. Model Eleven portable satellite/subwoofer/amplifier stereo
system, $749.00. Tested sample on loan from manufacturer.

Does this music system in a small suitcase belong in a
survey of loudspeakers? Well, where else does it belong? Its

11

pdf 12

salient qualities all have to do with the speaker designer's
art; electronically it's rather conventional. Designed by
Henry Kloss (a legend in his own time—or in his own
mind, depending on your perspective), the Model Eleven
represents some clever engineering in the size-versus­performance department. The so-called BassCase, a piece of
hard-sided luggage just under 20" long, holds all the pieces

—6½" high satellite speakers, 7" wide amplifier, all sorts of

cables and adapters, optional Walkman or Discman—and
when emptied becomes the airtight enclosure for the built-in

7" acoustic suspension woofer. Truly a virtuoso shoehorn

job. One little problem is that casual, sloppy repacking just

doesn't work; only the fastidious neatnik will be able to
close the repacked case. (There goes the youth market.)

I listened to the Model Eleven at some length, in the
company of several associates, before I measured it, and we
came to the conclusion that for an ultracompact trick system

it sounded remarkably complete and accurate but not quite

as good as the best conventional stereo systems of only

slightly larger size and comparable cost. For travel by auto­mobile, these other systems will fit into the trunk with equal
ease though perhaps not as neatly; for travel by plane, bus,
or train the Model Eleven is of course unbeatable. One thing
I faulted in the performance was an unpleasant shattering on
piano music; the little amplifier appeared to be clipping on
the peaks. On less dynamic music—cool jazz, for exam­ple—it's really a classy-sounding little system.

I must confess that I was unable to take this equip­ment as seriously as, say, a Snell speaker system, so that my
measurements were not very extensive. I did determine that
the BassCase woofer goes down to 42 Hz before starting to
roll off—very respectable for luggage. The electronic cross­over network incorporated in the amplifier is specced to op­erate at a crossover frequency of 150 Hz; from the response
of the satellites it looked more like 200 Hz to me, but even

150 Hz is a little high for completely nondirectional L+R
bass. The 3" midbass/midrange driver in the tiny satellites
appeared to be extremely flat in its range; the ¾" dome
tweeter, on the other hand, registered a very rough response
through the perforated metal grille (nonremovable and pos­sibly the sole cause of roughness). The crossover to the
tweeter is in the neighborhood of 4 kHz, as far as I could
tell by poking around in the nearfield. The amplifier can
also be powered from the 12-volt DC cigarette lighter sock­et of a car, but my aversion to elaborate auto sound kept me
from trying it that way. (I believe that one should be listen­ing to engine and road sounds with at least one ear when
driving and not be ecstatically plugged into Wagner or Jerry

Lee Lewis, oblivious to the audible world outside.) A 9-volt

DC power takeoff for your Walkman or Discman is on the

back of the amplifier.

As you can probably see from the above, I'm not the
right customer for the Cambridge SoundWorks Model Elev­en, but the right customers do exist, and I think they'll be
very pleased with the system. Henry Kloss still knows how
to juggle and massage the size/performance/price trade-off.

Carver "Amazing Loudspeaker"
Platinum Mark IV

Carver Corporation, P.O. Box 1237, Lynnwood, WA 98046. "The
Amazing Loudspeaker" Platinum Mark TV, $2199.00 the pair.
Tested samples on loan from manufacturer.

This unique loudspeaker has become something of an
obsession for Bob Carver, as indicated by the fact that the
Platinum Mark IV is its fifth-generation version (not count­ing the experimental versions that never went into produc­tion). On the back cover of Issue No. 15, a review of the
Mark III was announced as one of the coming attractions,
but Bob has meanwhile fiddled with the crossover network
and the frequency balance once again, so we're now look­ing at Mark IV. He says this is "It" now, no more changes,
but I'm skeptical. Not about the basic design, though.

I've said it before and I'll say it again: this is a classic,
a landmark design that rewrites the book in a number of re­spects. It's the first open-baffle loudspeaker system without
active equalization to come even close to state-of-the-art
bass performance. It's the first loudspeaker system to use a
monolithic ribbon-type line-source transducer successfully
all the way down to 100 Hz. (By successfully I mean with­out serious irregularities in response.) It's also the first gen­uinely clean large-signal loudspeaker system at anywhere
near its price. In other words, it's a breakthrough—in deed,
not just in claims. In a large listening room its dangerously
addictive qualities really assert themselves; I find even the
very best conventional, forward-firing enclosed speakers
somehow downsized and uninvolving by comparison, and I
soon go back to the Amazing. I'll never look at speakers in
the five-figure bracket the same way again because many of
them are simply not as good as this $2199 system.

Since this is my third review of Bob's brainchild—see
Issue No. 11 for my evaluation of the original version and
Issue No. 14 for the Platinum Mark II—I really don't want
to go over the same ground once again. New readers are
advised to obtain those back issues for a more thorough
discussion of the underlying design principles. Here I just
want to make a few additional comments and note the latest
changes.

I occasionally hear off-the-wall theoretical objections
to the Carver open-baffle bass system, which is the speak­er's most ingenious feature. You just have to listen to it, but
some people don't seem to trust what they hear or they
don't understand the concept. It's almost too simple to be
plausible. An open baffle necessarily creates a 6-dB-per­octave low-frequency decline. A woofer with an abnormally
high Q will have a big bass bump. The bump can be tailored
have a 6-dB-per-octave rise. Eureka! The two opposite
slopes will cancel out to create a flat response. In practice,
of course, it's not so simple. The shape and dimensions of
the open baffle, the woofer Q and resonant frequency, the
voice coil and cone must all be precisely designed and held

12

pdf 13

to tight tolerances, or the whole schmear just won't track as
a system. In fact, that's just what happens in planar speakers
with inerently high-Q bass panels—Magneplanar, Apogee,
etc.—which manage to produce a bass of sorts thanks to the

same laws of nature, but such a bass is not nearly as flat and
correctly damped as in the Carver because that "eureka"
perception of mirror-image slopes is not part of their design.

One of the pseudotechnical objection I've heard is
that, well, it's still a high-Q woofer, and we all know that
means underdamped. Wrong. The open baffle acts as an
acoustical short circuit, which lowers the Q analogously to
an electrical short circuit. Or, you could say that instead of
combining a conventional low-Q woofer with a convention­al high-Q box to produce the desired system Q, the Carver
accomplishes the same thing by combining a high-Q woofer
with a low-Q (and how!) open baffle. The difference is that
the response profile corresponding to the desired Q in the
Carver emerges only after the acoustical cancellation has
taken place, a small distance in front of the speaker; the ex­treme nearfield measurement still shows the high-Q bump

(obviously, the Don Keele method isn't applicable to open­baffle systems). Now, the Platinum version of the Amazing
has four 12" woofers per side, a total of eight, and the fun­damental resonance after break-in is in the neighborhood of

22 Hz, at which frequency the response is still essentially

flat. That combination of air-moving capability and low­frequency extension results in absolutely majestic, life-size
bass reproduction. No owner of the Amazing will ever need
to bring up the subject of subwoofers. In Mark III and Mark
IV, the acoustically derived equivalent Q is continuously
adjustable on the rear panel from 0.5 to 1.0 (no more resis­tors to insert, as in Mark II). Another change in Mark IV is
that the woofers and the ribbon all move forward in re-

sponse to a positive-going pulse, thus satisfying one of my

well-known little compulsions.

About those rear-panel controls—there are three of

them now and they work very nicely, but I disagree with the

way they are marked. The leftmost one is the Q control, and

its "recommended" start-up position is marked with a cali­bration line at 1.0 (all the way up clockwise). The flattest
bass response I measured in my large listening room was

obtained with the control at 0.7 (12 o'clock), confirming the

theoretical prediction. The 1.0 setting sounded too heavy.
The middle control has a range of approximately 6 dB for

adjusting the upper midrange, and its calibration mark is at

9 o'clock, whereas the measured flattest setting in my room

was at 3 o'clock. The far right control, with a similar range,

trims the high frequencies, and the calibration mark is again

all the way up clockwise. I had to back it off slightly to

about 4 o'clock for flattest response. I other words, the start-

up "recommendation" favors a shallow U-shaped frequency

response, with heavy bass, crispy highs, and a recessive
midrange—the kind of balance I generally associate with

unsophisticated hi-fi jockeys, who at the same time want to
be told that "everything is flat." Well, it's flat my way, not

their way. And here's what I mean by flat: I had to use my

Audio Control third-octave real-time spectrum analyzer in­stead of my trusty old Hewlett-Packard sweep spectrum an­alyzer because a 5-second log sweep is useless in the
farfield in a live room—and, as I indicated, this particular
speaker must be measured in the farfield—whereas pink
noise analyzed in real time still gives a more or less reason­able reading. Between the 1-dB-per-step and 2-dB scales I
estimated that the response was ±1.5 dB from 25 Hz to 20
kHz—that's the range of the instrument—at about 4 meters.
Not too shabby! Bob Carver claims that outdoors, where a
much more precise reading is obtainable, he can find a
sweet spot where the speaker is so flat, with the controls
trimmed in, that nobody would believe him if he published
the curve—but I believe him after my own measurements.
Of course, only the very best program material sounds best
when reproduced dead flat, but that's an old problem—and
its solution isn't a nonflat speaker.

Alvin Foster of the Boston Audio Society, with whose
perceptions about loudspeakers I nearly always agree, wrote
a 12½-page evaluation of the Amazing in The BAS Speaker
(Vol. 18, No. 1), covering various aspects of the subject in
much greater depth than I could ever hope to with my
review work load. I recommend this massive article—less
rigorous than an AES paper but more so than an underground
audiophile review—to all interested parties. (Address: The
BAS Speaker, P.O. Box 211, Boston, MA 02126-0002.)
Alvin confirms my previous findings as regards the large­signal capability, extended bass response, uncommonly low
distortion, and tremendous clarity of the speaker, but con­cludes that those are not the main reasons for its superior
sound. What then? He claims it's the dispersion or polar
pattern and "incredibly" flat overall frequency response.
That seems to contradict, at first blush, a recent mathemati­cal analysis of line sources by Stanley Lipshitz, who is not
in the habit of being wrong. Alvin has entered into a
dialogue with Dr. Lipshitz to try to find out why the Carver
ribbon has mysteriously better response than the mathemati­cal model would seem to permit. I'm sure there's a nonmys­tical explanation. I know, for example, that the ribbon is
passively equalized within the crossover/control network to
compensate for certain inherent acoustical radiation effects,
but there may be more to it than just that.

One minor annoyance I found in successive incar­nations of the Platinum version is a tendency to develop
very high-Q breakup resonances near the two ends of the
ribbon. These buzzes are heard only when the speaker is
swept with sine waves at a fairly high level; on music
there's no problem. Mark III was already very much cleaner
in this respect than Mark II, and in Mark IV the fault ap­pears to have been cured entirely. That ribbon is basically
nothing more than Reynolds Wrap glued to a plastic mem­brane, crinkled, and stretched on a frame between magnets;
it takes some production experience to keep it 100% stable.

I also want to emphasize again that (1) the Carver
speaker must be pulled well into the room, at least 3½ to 4
feet away from the back wall, to produce the kind of sound

13

pdf 14

I've been talking about and (2) it must be well broken in—
meaning about 50 hours of dynamic music—before it will
give you its absolute best performance. Your local dealer
probably won't have satisfied those conditions when you
ask to hear the speaker, and he probably won't be driving it
with over 500 clean watts per side as I do. I have no control
over his particular situation or yours; I'm just telling you
what happens in my listening room.

Let me conclude with something I may not have fully
communicated before. Consider the most exalted ultrahigh­end loudspeaker systems in the world: the Infinity IRS and
IRS Beta, the top-of-the-line Martin-Logans, the Sound Lab
A series, the Wilson Audio WAMM, the Thiel CS5, the
Duntech Sovereign, the Apogee Diva, the new B&W Ma­trix 800, and others in that general bracket. Don't for a
moment imagine that any of them is strikingly or over­whelmingly better than the Carver "Amazing Loudspeaker"
Platinum Mark IV, that stepping up to one of them from the
Carver is to step into another world. No way. I'm not about
to make comparisons here; you may end up preferring this

one or that one for various reasons—or even the Carver
over all of them, for the reasons already discussed. The
point is that the Carver is right up there—almost as good,
better by a hair, not really as good, or what have you, but in
there—at $2199 the pair! That constitutes a very serious
political problem which the high-end community isn't ready
to deal with. By solving certain long-standing technical
problems in a dramatically cost-effective way, Bob Carver
has created as much of a monster for the industry as a boon
for audiophiles. History will end up being on his side, but
for the moment the high-end community is not. I've always
enjoyed the spectacle of second-rate minds freaking out
over a first-rate reality, so I'm having fun with the Carver
monkey wrench in the high-end mystique, but sooner or lat­er that reality will have to be accepted by all rational audio
people. Don't wait until then, however, to check out the
Carver speaker and form your own opinion. Life is short.

JBL XPL160A

JBL Consumer Products, Inc., a Harman International Company,
240 Crossways Park West, Woodbury, Long Island, NY 11797.
XPLI60A floor-standing 3-way loudspeaker system, $2498.00 the
pair. Tested samples on loan from manufacturer.

A major paradox of the loudspeaker industry: JBL

makes the best drivers, has the slickest production tech-

niques, and is both progressive and honest in the R & D

area—yet there seems to be no truly first-class JBL speaker
system for home use (as distinct from professional sound).

The XPL series is supposed to be JBL's breakthrough in the

audiophile market, but on the basis of the XPL 160A I can't

confirm that. It's a frustrating, self-contradictory speaker.

In my review of the JBL L40t3 two issues ago, I

called the proprietary pure-titanium 1" dome tweeter in that

system the best known to me, bar none. I have to reiterate

that opinion now, after having tested an updated version of
the same tweeter in the XPL160A. And that's not all. The
midrange driver in the XPL160A is designed around a pure­titanium 3" dome, a tour de force never before attempted to
my knowledge, certainly not as successfully as in this re­markable unit. How they got rid of all the standing waves is
beyond me, but they did. The two dome drivers are mount­ed as close together as possible and crossed over at approxi­mately 4 kHz to form what functions, in effect, as a single
seamless 1 kHz to 20 kH transducer of the utmost flatness
and smoothness. The two voice coils are wired out of phase,
probably as a concomitant of a second-order network.
There's no high-frequency peak; the rolloff begins at 20
kHz but stops and reverses a bit after 30 kHz. The older ver­sion of the tweeter went out a few more kHz on axis but
wasn't quite as well damped, and the off-axis rolloff began
sooner. The double-dome combination has just about the
same response 30° off axis as on axis, meaning almost per­fectly flat (when the microphone axis is at the most favor­able height) and showing much smaller squiggles—maybe 2
dB from peak to peak—than I've seen in other drivers. No
trace of ringing of any kind, either. If I were in charge of a

new project to design a conventional electrodynamic speak­er system, these are the drivers I'd like to specify because
they're simply the best; unfortunately JBL keeps them
strictly in-house.

So far so good—indeed, fantastic. The 10" woofer in
its vented box is another matter. The 33" high cabinet itself
is gorgeous—high-gloss black lacquer finish, subtly nonpar­allel side walls (trapezoidal cross section), neoprene-lined
baffle step to time-delay the domes (very impressive crafts­manship), elaborate open grille frame, and so forth—but the
tuning of the woofer enclosure appears to be less than opti­mal. The box frequency (where the displacement of the
woofer cone is at a minimum) is 34 Hz, but maximum out­put from the rearward-facing vent is at 44 Hz, and that
looks like the effective low-frequency cutoff of the system.
I've seen deeper bass out of smaller boxes at the same
efficiency (between 88 and 89 dB SPL at 1 meter with 1
watt input).

The paper cone of the woofer is, as far as I can tell,
the downfall of this speaker system. Its frequency response
is extremely flat, but there are—you guessed it—energy
storage problems. Tone burst tests revealed definite ringing
in the octave just above the crossover frequency of 800 Hz,
where the 12 dB per octave rolloff begins. (The 3" dome
actually comes in just above 1 kHz, also with a 12 dB per
octave slope, but for some reason there's no hole in the
summed response, perhaps because the woofer and mid­range are wired in phase despite the second-order cross­over.) If the otherwise excellent woofer were used only up
to, say, 400 Hz, there would be no problem, but with the
800 Hz crossover the ringing in the 1 kHz to 1.5 kHz range
is insufficiently attenuated and becomes the signature of the
speaker. I have a feeling that a you-must-use-what-we-have
corporate policy was imposed on the engineers here. For all

14

pdf 15

I know, the woofer was conceived for a totally different

application; it's very good, for example, in terms of linear

excursion—the Q doesn't change at all as the amplitude of a

step-function input is increased.

The sound that results from this mixed bag of design

elements is intriguing but ultimately unsatisfactory. Above

2 kHz or so everything is utterly neutral, transparent, and
smooth as silk, as good as you'll ever get out of a forward­firing system. The barely attenuated ringing immediately

above the woofer's passband, however, is a pervasive color­ation at all times and on all types of music. It comes off as a
breathy hollowness, and it's right there in the midrange
where you can't get away from it. I find it to be a disqualify­ing fault of what would otherwise be a stupendous-sounding
loudspeaker, albeit somewhat light on the bottom end. The
top-of-the-line XPL200, which has been promised to me for
review, has a 12" woofer and a separate lower-midrange
driver covering the two octaves from 300 Hz to 1.1 kHz, so
it probably avoids the same pitfall. Those titanium-dome
drivers deserve the best possible system design.

Snell Type C/IV

Snell Acoustics, Inc., 143 Essex Street, Haverhill, MA 01832. Type

C/IV floor-standing 3-way loudspeaker system, $2190.00 the pair.

Tested samples on loan from manufacturer.

In Issue No. 13, I wrote that "the Snell Type C/II is
just about a state-of-the-art 'monkey coffin' (trade slang for
a conventional forward-firing one-piece speaker system in a
rectangular box)," and in Issue No. 14 I followed that up
with a fairly detailed review explaining why I think so.
Type C/IV is a successor model in exactly the same format,
so I won't start at square one here to describe the system;
you may want to refer back to the C/II review. (In case you
wonder about Type C/III, it was discontinued almost as
soon as it was announced.)

So now the Snell Type C/IV is the state-of-the-art
monkey coffin. Yes, it's better than the C/II in a number of
ways. The bass is greatly improved; I measured a classic
fourth-order Butterworth response—vent response peak pre­cisely filling in the woofer null at a box frequency of 24 or
25 Hz. That's quite an achievement with a 10" woofer, an
internal volume of less than 3½ cubic feet (estimated), and
fairly high efficiency (88.5 dB). I'd say Kevin Voecks is
"pushing the envelope," as the saying goes in high-tech
country. He wasn't when he did the bottom end of the C/II.
The large-signal step response of the C/IV woofer is a bit
more Q-ey than the small-signal step response, indicating
less than perfect linearity on long excursions, but nothing in
this world is perfect. It's still a very impressive 10" bass
system.

The front tweeter of the C/IV is also new and ostensi­bly improved, although the old soft-dome unit was certainly
good enough. The Vifa metal-dome tweeter now favored by

Snell is claimed to be the best representative of the breed; I
find that JBL's proprietary titanium dome is even better, but
the Vifa is indeed extremely flat and smooth in response. It

begins to roll off at 18 kHz, then kicks up again and comes
to a high-Q peak at 25 kHz. That's fairly typical of metal
domes and completely unobjectionable to me. (I don't know
how my dogs feel about it because I keep them out of the
laboratory.) Front tweeter level is continuously variable; the
calibrated Optimal position of the level control appears to
be accurate. The rearward-firing little Audax supertweeter
(for "air and balance") remains unchanged, as is the on/off
switch for it. And, yes, the two pairs of terminals for bi­wiring are still there, in genuflection to unscientific cultism
by an otherwise scientific company, but—what the hell—
they do no harm.

My measurements clearly indicated that the frequency
response of the total system is optimized/normalized to the
axis of the midrange driver, where the deviation from abso­lute flatness is no more than ±2.5 dB, maybe only ±2 dB.
That's for the full range from deepest bottom to tip-top—
truly remarkable. Off-axis response remains almost as flat
over an impressively wide angle. (See also David Rich's re­view of the Snell Type Q in this issue for his observations
about Snell's design approach, their QC procedures, and
their use of Floyd Toole's NRC facilities in Canada.) Tone
bursts swept through a wide range of frequencies revealed
negligible storage. Woofer, midrange, and tweeter are con­nected in phase—a positive-going pulse makes them all
move forward—but a square pulse input cannot be acousti­cally recovered from the speaker regardless of where the
microphone is placed. That, of course, is the nature of the
beast—a 3-way system with 4th-order Linkwitz-Riley
crossovers—and Snell has never considered pulse coher­ence to be a design requirement. (This is not the time and
the place for a dissertation on the audibility or inaudibility
of phase.)

The sound of the Snell Type C/IV is, yes, the best I've
ever heard out of a monkey coffin—uncolored, transparent,
low in distortion, high in resolution, perfectly balanced,
much better on the bottom end than the C/II. The frequency
balance—which depends not so much on whether the re­sponse is ±2 dB, or ±2.5 dB, or whatever, but on just where
those little zigs and zags occur—is probably the most satis­fying of any speaker known to me. To be sure, there's more
to a speaker than frequency balance—for example, the
Carver "Amazing" at exactly the same price produces a
larger, more authoritative, more dynamic, more concert­hall-like sound—but even the Carver could use the C/IV as
a model in the frequency balance department.

To say something negative—I have to search for it—
the cabinet quality could be a little higher considering the
price, not so much in basic construction but in little details
of finish. I have a feeling that the C/IV is somewhat costlier
to make than the C/II, and something had to give. Don't let
that stop you from giving very serious consideration to this
outstanding loudspeaker system.

15

pdf 16

Velodyne ULD-15 Series II

Velodyne Acoustics, Inc., 1746 Junction Avenue, San Jose, CA

95112. ULD-15 Series II subwoofer system (single unit) with pow­er servo controller, $1795.00. Optional passive highpass/bypass

accessory available. Tested samples on loan from manufacturer.

This subwoofer system has been around for quite a
few years, but I never had a chance to get my hands on one.
More recently its designer formalized his ideas on the tech­nical aspects of the subject in an Audio Engineering Society
paper (David S. Hall, "Design Considerations for an Accel­erometer-Based Dynamic Loudspeaker Motional Feedback
System," 87th Convention of the AES, New York, NY, 18­21 October 1989, Preprint 2863), which rekindled my inter­est and motivated me to start bugging the company for a
review sample until I got one. What they sent me was the
Series II update of the original 15-inch model (12-inch and

18-inch versions also exist).

My curiosity about the Velodyne must now be bal­anced against my well-known—or at least frequently
avowed—reluctance to go over the same ground that some­one else has already covered with great competence. I'm
referring to David L. Clark's exhaustive and authoritative
six-page review of the original ULD-15 in the November

1987 issue of Audio, free reprints of which are available
from Velodyne. There's really very little I could say about
the almost identical Series II that DLC hasn't explained in
considerable detail. The difference appears to be mainly a
beefed-up version of the external power servo controller,
which now has a rated amplifier power output of 400 watts
rms and incorporates some circuit changes. Between the
Hall paper and the Clark review, there's no opening left for

piercing insights by your Editor. Even so, I must attempt an

appreciation (in the literary sense), a full-fledged technical
test report being clearly unnecessary.

The system is probably the most highly refined em­bodiment of the motional-feedback approach to bass trans­ducer design, the principal advantage of which is greatly
reduced harmonic and intermodulation distortion. The Hall
paper claims an improvement by a factor of 10 over conven-

tional woofers, and the Clark review confirms that. It should
be noted, however, that conventional woofer distortion is
largely excursion-related, so that the Carver "Amazing
Loudspeaker," for example, with its four 12" woofers per

side achieves comparably low distortion figures simply by
dividing up the total excursion requirement among a larger
number of feedbackless drivers. The uniqueness of the Velo­dyne is that it allows almost any reasonably good pair of

speakers to acquire ultralow-distortion bass, flat all the way
down to the limits of hearing, and takes up only 2½ square
feet of additional floor space. The ULD-15 is normally de­livered with the active lowpass and highpass filters in the
controller unit set to 12 dB per octave slopes and a nominal
crossover frequency of 85 Hz. Other frequencies, as well as
6 dB per octave slopes, are available as a dealer-installed

option. A further option is the passive highpass/bypass
switching box, which allows the main speakers to be
crossed over passively with 6 dB per octave slopes or to be
played full range without the subwoofer. Thus there exists
more than the usual degree of flexibility in the main-to-sub
marriage, although David Rich's caveats on that subject
(see the article that follows) still apply.

I was particularly interested in how the ULD-15
would complement the Quad ESL-63. That's one great
speaker that can definitely use bass extension to live up to
its full potential. The default frequency of 85 Hz for the
crossover point seemed about right, as it overlaps the bot­tom end of the ESL-63 by an octave, eliminating the need
for sophisticated matching and allowing confident use of the
active 12 dB per octave lowpass and highpass sections. The
audible results were excellent, even with just one ULD-15,
although I would have preferred two. Directionality is not
an issue below 85 Hz, but room excitation at two points, 3
dB down each, will produce a less aggressive complex of
standing waves than single-point excitation at full power;
furthermore, two subwoofers provide 3 dB more headroom
on bass transients and generally tend to give a more com-

plete impression of the low-frequency characteristics of the
concert hall. Velodyne, however, is promoting the mono­lithic matrixed subwoofer concept, so reviewers get one unit
and that's that. At any rate, the transition between a pair of
Quads and a single Velodyne appeared to be quite seamless
to me—yes, tweaks, the ULD-15 is "fast" enough for the
electrostatics, whatever that means. (There's no such thing
as a fast woofer, boys and girls. If a woofer were fast, it
would be a tweeter. I think semieducated audiophiles mean
a well-damped woofer without hangover when they use the
word. Motional feedback certainly achieves that.) Despite
the smoothly and deeply extended bottom end, the Quads
still don't sound like big speakers with lots of headroom.
They sound like Quads with deep, clean, detailed bass.
That's far from the worst thing that can happen to a music
lover, to be sure, but a large-signal Quad is not yet a reality.

Level matching to the main speakers is a sine qua non

with the Velodyne, although at the CES and in other com­mercial demonstrations the level is always cranked up to
show off the amazing bass, so everything sounds thick and
unnatural. With the 85 Hz crossover and 12 dB per octave
slopes, proper level matching means that you'll hear no dif-

ference at all on certain kinds of music when the subwoofer

is bypassed and the main speakers are allowed to play full
range. Only when there's real bass should you hear any, but

then you should hear it life-size and perfectly defined. To
obtain that kind of correct level adjustment in a real-world
listening room, your ears are probably the best instrument,

especially if you move around the room and experiment

with many different recordings. Remember that bass is the
foundation of music but not its sole purpose.

Of course, you're probably aware of the audiophiles
whose taste is so exquisite that they regard all bass as vulgar
and unnecessary. The Velodyne is definitely not for them. 0

16

pdf 17