The answer to this
question, in my opinion, has to start with an overview of quite an
interesting phenomenon that is responsible for appearance and
dissemination of superstitions and myths in High End Audio. I'm
talking about a specific conception that, by now, has become deeply
rooted in the minds of audiophiles and many professionals. This
conception promotes the absence of correlation between the
objectively measured characteristics of the sound reproduction
equipment and the results of subjective listening.
In order to save some
space, I will limit my discussion to amplifiers, although everything
I say below applies to any audio component.
The tenacity of these
myths has to do with consistent adhearing to the-now-traditional
methods of audio system evaluation, methods that use generally
accepted formal parameters from product specifications. This is
particularly obvious in reviews in specialty audio magazines that do
both the objective and subjective tests, and that echo the
traditional approach to evaluating a component. There are plenty of
reviews on certain amplifiers which are, even by the least demanding
standards, not very good in the opinion of test engineers (who do
the objective measurements) but which are sometimes praised to no
end by the reviewer (who does the listening part of the review), and
vise versa.
Such discrepancies
between measurement results and listening evaluation are the direct
result of a lack in the field of audio electronics of objective
criteria that would allow us to evaluate the component sound quality
without auditioning it.
I will try to clarify
the essence of the above issue in a context of my research conducted
over the years and within the limits of the "human hearing
mechanism" theory which I developed in the course of this work.
I apologize for the gross oversimplification in describing the
principles at the basis of my design philosophy and their
application because the mathematical apparatus involved is quite
complicated and is well beyond the scope of this answer. What
matters here is the person's understanding of the real and objective
tool for evaluating the sound quality, not techno-babble behind the
tool. In this oversimplified description I will try not to throw the
baby out with the bath, so to speak.
In my opinion, one of
the key steps that would bring us closer to the objective evaluation
of the measured parameters/sound quality of the amplification gear
and the establishment of unambiguous correlation between the two is
to introduce a new notion that would allows us to interpret the
measurement results and predict the sound quality of the measured
piece of equipment without listening. I take the liberty of labeling
such a notion the "Absolute Linearity of a System" (ALS)
because no commonly used terminology exists for describing this
concept. Since, in general, the ALS is a function of a certain
number of variables, explaining the influence and interaction of all
of these variables is not realistic within the limits of the answer.
Therefore, for the purpose of clarifying one of the aspects of ALS,
I'll concentrate on the interaction of only two variables which are
commonly used for component evaluation, and to which much
significance is ascribed in the technical specs of any amplifier.
These two variables are also a subject of most diverse
interpretations in comparing subjective (listening) and objective
(measurement) results. Keep in mind that we take these two variables
out of context, and ONLY for the purpose of demonstrating certain
aspects of the ALS. These two variables are:
-
TOTAL HARMONIC
DISTORTION (THD) vs. FREQUENCY
at various power levels and
-
THD vs POWER (usually,
these measurements are taken at 1KHz, although it is necessary to
take them at least at three frequencies: 20Hz, 1KHz and 20KHz, or
sometimes at 10KHz instead of 20KHz)
Figure 1 (below) is a
graphical expression of THD vs. FREQUENCY of a so-called ideal
amplifier (ideal in a sense of
faithfulness of sound reproduction).1
As you can see, this parameter is a straight line at each power
level; it is constant and does not change with frequency.
Figure 2 (below) is a
graphical expression of THD vs. power of the same so-called ideal
amplifier.1
As you can see, this curve has the same shape at all frequencies
within the audio frequency range; THD smoothly increases with an
increase in power.2
(Description of the behavior of this curve lies beyond the scope
of this answer.)
[1
Again, a strict proof of this fact is beyond the scope of answering
this question.
2
In a so-called "ideal" amplifier all three curves in Fig.2
must merge into one line, in other words, they must be absolutely
identical. I spread them out for easier viewing.]
Both of these curves (in
Fig. 1 and Fig. 2) are, of course, part of the three-dimensional
representation of the function of two variables, which somehow can
be expressed in this manner:
THD=Function (power,
frequency)
If the corresponding
characteristics of a real-world amplifier are close to those
described above, then the combination of THD vs. FREQUENCY, THD vs.
POWER and some other parameters become the variables of another,
more refined, system of parameters. This higher-level system
absolutely unambiguously describes the sound quality of any
amplifier based on its formal parameters as they appear in
manufacturer's specs.
As mentioned earlier, it
is not feasible here to review all variables. However, in addition
to information about THD vs. FREQUENCY and THD vs. POWER, it is
absolutely necessary to at least have information about the HARMONIC
DISTORTION RESIDUE (HDR) at various frequencies within the audio
frequency range and at various power levels. Keep in mind that very
rigid limitations apply to the form of the HDR, limitations which
are imposed on the number of harmonics and the ratio between the
values of different harmonics.
Once these requirements
are met, THD vs. FREQUENCY, THD vs. POWER, information about HDR and
some other parameters3
become objective tools in evaluating the sound characteristics of an
amplifier. We have to keep in mind that we should look not as much
for the value of these parameters per se4
but rather for their graphical or analytical expression because the
value itself does not carry information which is traditionally
ascribed to it. What matters is the difference between the behavior
of a so-called ideal system and the system under evaluation. Again,
the "Absolute Linearity of a System" is the ability of an
amplifier to exhibit parameters close to those of an ideal
amplifier.
[3
Those presented in an amplifier's specs and which, in the way they
are presented, mean very little and are often confusing in their
correlation with the sound.
4
The value of these parameters, in general, are more applicable to
clarifying purely technical aspects, such as evaluation of the
amplifier's topology, and pointing out the types of active elements
utilized (solid-state devices or vacuum tubes), etc. Of course,
these data provide certain information about the quality of sound.
However, when taken out of context and being incomplete, these same
data is what brings about the well-known "paradoxes" in
comparisons of the objective vs. the subjective.]
What's interesting is
that if a hypothetical engineer were to implement these principles,
he would sooner or later arrive at a limited number of topologies
conforming to the described requirements. On the way to such a goal,
this engineer would have to discard many conventional topologies,
some of which are quite beautiful and appealing in layout.
Please note that when
one tries to implement these methods into the amplifier design without
any tricks, one realizes that the cost of
production of such a unit is quite high. Which is the case,
of course, with all real things in this world.
From my point of view,
utilization of these methods sets very high standards in equipment
design and, in turn, leads to a increase in production cost
which determines a correspondingly high retail price. At
the same time, we are able to hear immediately the difference
between the system based on my design principles and other systems.
LAMM equipment
represents a life-time investment. It actually saves one a lot of
money in the long run by eliminating the need to constantly upgrade
one's system.
To corroborate my point
and to prove that I have been able (to a greater or lesser degree)
to implement the principles described above, I have attached the
FILES
containing the results of objective measurements of certain
parameters for one of my designs, single-ended amplifier model ML2.
All these measurements were conducted by an independent party, Mr.
Bascom H. King of BHK Labs (bhk@rain.org).
I picked this amplifier as an example on purpose because it is quite
expensive and because it is a single-ended design which
traditionally measures very poorly.
The ML2 has already been
reviewed by four different magazines (click
here for a list of reviews).
In the future I will
continue the discussion of this issue with the purpose of showing
the unambiguous correlation between the methods I use in designing
the ML2 (and other models), measurement results, sound quality, and
the price that I have to pay to attain various results in various
models...
© 2004, Vladimir Lamm.
All rights reserved.
M1.2
Reference
