Kevin M. Hayes
Valve Amplification Company
As published in Positive Feedback magazine
Introduction
In the 4th century BC Solomon wrote, "There is nothing new under the
sun." Of course, he was referring to the human heart, but the same is
largely true of high-end audio as well. In the past twenty years speaker
manufacturers "discovered" time alignment and diffraction
control, yet these topics are covered in the 1952 Radiotron Designers'
Handbook. Single-ended tube power amplifiers are gaining a reputation for
lush sound due to double-digit quantities of second harmonic distortion,
but this trick was used in budget radios in the 1940s. The well known "ultra-linear"
output circuit is most often attributed to David Hafler and Herbert
Keroes, but was patented in England fifteen years earlier by Alan
Blumlein.
Most audiophiles (and some designers) do not see the audio art in its
historical dimension. Yet the context of history is of great value in
assessing the current position of our pursuit. An improved understanding
could even assist audiophiles as they consider purchases of new equipment.
Toward this end, we will set forth a framework for viewing history, and
then illustrate with a brief consideration of two topics: the debate
between tubed verses solid-state electronics, and the stability of the
digital state-of-the-art.
Innovation
In the book Innovation, author Richard Foster describes a phenomenon
displayed by almost all products and technologies called the "S-curve."
In Figure 1, the vertical axis is performance (or economy) and the
horizontal axis is effort invested (time, money, etc.). These two
parameters are related by an "S" shaped curve. This curve
reflects the fact that it takes a lot of effort to get an idea to work at
all (the first flat bit), and then suddenly everyone sees how it basically
works and can think of lots of ways to improve it (the vertical rise).
After a while it's hard to think of ways to make a meaningful and cost
effective improvement (the last flat bit), and further effort is rewarded
with diminishing returns.
Most new ideas or products conform to the S-curve. Think about pocket
calculators over the last twenty years and you'll get the picture.
What happens when two technologies exist to perform the same function?
Well, each product has its own S-curve, and they can be shown on the same
graph.
When a new technology debuts in a commercial embodiment, it is typically
somewhere near the knee of the curve leading into the vertical rise. At
the same time, the older technology is undoubtedly much further along its
curve, and may even have topped out. For example, Figure 2 is a
hypothetical comparison between vacuum tubes, discrete transistors, and
integrated circuits for use in digital computers. Early transistors had
unexplained and unexpected modes of failure, but that got sorted out.
However, integrated circuits allowed for faster cycle times and vastly
reduced assembly labor, resulting today in a palm-top calculator that can
easily outclass the Air Force's Whirlwind machine of the late fifties.
It is interesting to note that the newer technology usually becomes
dominant before its performance equals the more mature technology. It
takes over based on the perceived possibilities for the future.
Tube vs. Transistor
Sight and hearing are what a natural scientist might call "touchy-feely"
areas. By definition, both only exist as perceived by an organism. In the
case of humans, sound is analyzed by a tremendously complex and not well
understood series of operations that can be loosely lumped together and
labelled "pattern recognition." The important point here is that
no test bench in the world can come close to listening to things the way
humans do. I will provide further musings about the test bench in a future
article, but for now I will simply wave my hands and assert that Figure 2
is not applicable to audio equipment.
Figure 3 illustrates where I believe we are with audio electronics. The
transistor began with a great deal of promise, but the first units were
noisy, unreliable, and in some cases wore out faster than tubes (yes,
transistors do wear). Still, some transistor based audio electronics were
marketed in the 1950's. By the mid 1960's it was possible to make a
reliable transistor amplifier, but they still sounded awful. In the
1970's, certain important ideas about the topology of transistor
amplifiers were worked out. In recent years, however, there have been very
few really new insights into how to make a solid state amplifier, and the
sonic state of the art has become rather static.
In fact, improvements in the sound of tubed and transistorized equipment
is now moving at about the same pace, and probably relates primarily to
factors common to both technologies, including power supply concepts,
incremental improvements in passive parts, and less reliance on
traditional test measurements (again, more next time).
This situation is annoying to many solid-state protagonists. He discovers
a transistor amplifier that sounds almost as good as tubes, only to
discover that tube amplifiers have also gotten a bit better. This state of
affairs can be understood by recognizing that both technologies are now
very mature. Indeed, semiconductor technology (in the form of a diode)
dates back to 1835, long before the earliest tube devices. Since the
announcement of the transistor in the late forties, huge amounts of money
and R&D have gotten us to where we are today. Both tubes and
transistors are in the flat bit at the top of their S-curves, and the
differences you hear now between the technologies will probably still be
around for your grandchildren to hear, unless something totally new
appears.
In an interesting footnote, researchers at MIT's Lincoln Laboratory have
developed a new cathode material for vacuum tubes that emits a reasonable
stream of electrons at room temperature. Image this developed into, say, a
12AX7 without a heater. A preamp built with this could be as cool,
long-lived, and compact as a solid-state unit. According to one
researcher, subminiature tubes could be developed that have higher power
and frequency capabilities than even diamond transistors with virtually
unlimited life and reliability. This is one instance where I miss the cold
war: tubes are far less sensitive to radiation than semiconductors, and
the military probably would have funded development of the thing. The new
cathode material qualifies as a new technology, and so should show its own
S-curve distinct from the thermionic vacuum tubes now in use.
Wither Digital?
Where do the S-curves for analogue and digital meet? It's hard to say
with certainly, but I have noticed a real levelling off of PCM digital
(CD) innovation in the last few years. What we see now are little
tweekings of the major decoding schemes (Bitstream and ladder-DAC) and
lots of marketing hype. In truth, I think there are a few D/A converters
on the market today which will still sound competitive in ten years. The
big developments are not going the be in the sound of the very best units,
but in driving the drek out of the lower price ranges. Thus, the best
units are near the top of the S-curve, but it will take a while for most
of the field to catch up.
For the audiophile, this means that there are a very few $3000+ D/A
converters that can be purchased with confidence. Listen very carefully,
though, in comparison to the best analogue or recent memory of a live
acoustic performance. Below $3000 there will be a lot of sonic changes
coming.
As stated earlier, the newer technology tends to become dominant long
before its performance matches that of the older technology, and this was
certainly true for compact discs. It is only now that vinyl is all but
dead that CDs have reached true fidelity. For the audiophile, the
intervening period was pure torture.
This discussion applies to the current PCM compact disc format, which is
frozen for all intents and purposes. The nature of future digital formats
will be determined by the outcomes of battles now underway between those
who want to see more information captured (i.e., more bits and/or faster
stored sample rates), and those who actually want to store less
information. An example of the latter is the new DCC format, which tosses
away information that the designers' models predict you can't hear. It is
truly unfortunate that such models are invariably based on the concept of
a "normal" listener, and discounts the keenest ears and the most
impassioned listeners.
Parting Shots
There is a final lesson to be taken from the study of innovation, which
might be called the "white elephant" phenomenon. As an old
technology is being eclipsed by a new one, someone will invariably throw
in the kitchen sink in an attempt to do better with the old technology.
Some unsung ship builder tried to make a king-size sailing freighter to
defend against the new steamship, and it broke apart in a spectacular way.
John Henry drove himself into the ground to beat the steam hammer. For
tubes, the white elephant was the nuvistor. For LPs it was probably the
Finial turntable.
Occasionally one sees the combination of competing technologies, often
resulting in the worst of both worlds. A freighter with both sails and
engine presents conflicting needs for hull design, strength, and crew. An
engine that can fire on two, four, or eight cylinders is usually a mess.
There may be reason to view with suspicion such things as hybrid
amplifiers, dacs combining ladder and Bitstream techniques, and digitally
remastered LPs.
Almost every technology has an appropriate application. For a computer or
a portable telephone, you can't beat the transistor. For no-holds-barred
high fidelity audio, you can't best the vacuum tube.
