(Audio Guy) wrote on r.a.h-e in message
...
Audio is a trivial application, they learn about power supply
design and amplification, which is pretty much all there is to
audio amplifiers, in their early years and then go on to much more
interesting and challenging concepts.

I see statements like this from time to time, yet I am not so sure
that audio design is "trivial." There are not many other design fields
where an amplifier: has to provide up to 30dB of voltage gain; act as a
voltage source into a wide and arbitrary range of load impedances
and do so in an unconditionally stable manner; have a passband noise
contribution at least 90dB down from 1W into 8 ohms, no matter what
its voltage gain and ultimate power delivery; have distortion
components under all load conditions that are below the threshold of
hearing no matter what the program material is; and do all the above
over at least three-decade, ie, a 10-octave passband.

Thoughts, gentlemen? I would suggest that designing, say, a typical RF
amplifier is, by comparison, "trivial" but, of course, I may just be
missing something :-)

John Atkinson
Editor, Stereophile

"John Atkinson" wrote in message
om...
(Audio Guy) wrote on r.a.h-e in message
...
Audio is a trivial application, they learn about power supply
design and amplification, which is pretty much all there is to
audio amplifiers, in their early years and then go on to much more
interesting and challenging concepts.

I see statements like this from time to time, yet I am not so sure
that audio design is "trivial." There are not many other design fields
where an amplifier: has to provide up to 30dB of voltage gain; act as a
voltage source into a wide and arbitrary range of load impedances
and do so in an unconditionally stable manner; have a passband noise
contribution at least 90dB down from 1W into 8 ohms, no matter what
its voltage gain and ultimate power delivery; have distortion
components under all load conditions that are below the threshold of
hearing no matter what the program material is; and do all the above
over at least three-decade, ie, a 10-octave passband.

Thoughts, gentlemen? I would suggest that designing, say, a typical RF
amplifier is, by comparison, "trivial" but, of course, I may just be
missing something :-)

John Atkinson
Editor, Stereophile

I don't think it's trivial at all.
It does, however, tend to be canonical, which means the devil is in the
details, rather than a new overarching concept which probably does not
exist.

It might be interesting to enumerate the design concepts for solid state
amplifiers which have occurred. I would guess the number to be less than
twenty, in two groups:
1. Device physics
2. Circuit topology

From the examples I've seen, written about primarily in Stereophile --
though the Acoustat was covered in "Audio", it would seem that the result is
limited more by the design budget than anything else. The more components,
as in active constant current sources, higher quality parts, stiffer
supplies, etc., one can throw at it, the better the result. The end game
example of this is that Australian amp (name, please?) that has far lower
levels of distortion than anything else.

That amplifier is an important example, because it tends to negate the worth
of the "boutique designs" that emphasize some particular parameter at the
expense of others. That is another area that I've never found real happiness
with, in comparison to amplifiers engineered for general goodness.

Oddly, although amplifier design has become mature, it does not appear to me
that specification and testing has. The scientist in me says it all has to
reduce to numbers, yet it does not, which means that the numbers are
obtained by test procedures that fail to characterize amplifier behavior.

Someone ought to work on this.

In article ,
Robert Morein wrote:

supplies, etc., one can throw at it, the better the result. The end game
example of this is that Australian amp (name, please?) that has far lower
levels of distortion than anything else.

Lower levels of distortion is not
the end game.
Stability into infinitely varying loads,
is.

Oddly, although amplifier design has become mature, it does not appear to me
that specification and testing has. The scientist in me says it all has to
reduce to numbers, yet it does not, which means that the numbers are
obtained by test procedures that fail to characterize amplifier behavior.

Someone ought to work on this.


Audio amplifier design is fascinating, and it's
what drew me to rao at first. Early discussions on
Class A vs A/AB design, and 300B tube designs
were a great education for me.

Of course, then, those with knowledge left, leaving
the wreckage you see.

It's great to see interest in improving audio
design.


Joe

"Joe Duffy" wrote in message


Of course, then, those with knowledge left, leaving
the wreckage you see.

Not everybody with knowledge left.

It's great to see interest in improving audio
design.

The most significant gains are needed elsewhere. Amp technology isn't
trivial, but it is pretty well cut-and-dried. There's not much left to do
but to make them smaller/cheaper/lighter/more efficient.

In article ,
Arny Krueger wrote:

The most significant gains are needed elsewhere. Amp technology isn't


I agree that speakers yield the most improvement,
however amplifiers are more interesting.


Joe

The "perfect amp" I cited is Halcro.

If science is to be believed, there can be no competitor to this company's
product.

"Robert Morein" wrote in message

The "perfect amp" I cited is Halcro.

If science is to be believed, there can be no competitor to this
company's product.

If science is to believed, Halcro levels of technical perfection are an
interesting technical exercise and little else.

"Arny Krueger" wrote in message
...
"Robert Morein" wrote in message

The "perfect amp" I cited is Halcro.

If science is to be believed, there can be no competitor to this
company's product.

If science is to believed, Halcro levels of technical perfection are an
interesting technical exercise and little else.

I distinguish between soft science, which encompasses psychology,
perception, and certain aspects of biology, and hard science, such as
physics and molecular biology.

Because soft science relies so much on observation, without any overarching
theory to back it up, it is prone to error and reversal. For example, for a
hundred years, it was believed that neuronal replacement did not occur in
the adult mammalian brain. Within the past three years, this has become
known to be completely false. And the ear is an extension of the nervous
system.

Your statement above may be correct, or it may not, but there is no good
science to back it up. All there is are isolated studies, flawed or not,
which may be used to incorrectly extrapolate the conclusion that perfection
at the level of a Halcro is irrelevant.

Bad scientist alert STANDS!!!

"Robert Morein" wrote in message

"Arny Krueger" wrote in message
...
"Robert Morein" wrote in message

The "perfect amp" I cited is Halcro.

If science is to be believed, there can be no competitor to this
company's product.

If science is to believed, Halcro levels of technical perfection are
an interesting technical exercise and little else.

I distinguish between soft science, which encompasses psychology,
perception, and certain aspects of biology, and hard science, such as
physics and molecular biology.

I think most people do. But just because there's a distinction in our minds
doesn't mean that the soft sciences are 100% bad which you later claim.
Furthermore, you hedged your bets by putting a undefined boundary line out
someplace in the science of biology. So you admit that its not clear where
this dividing line is.

Because soft science relies so much on observation, without any
overarching theory to back it up, it is prone to error and reversal.
For example, for a hundred years, it was believed that neuronal
replacement did not occur in the adult mammalian brain. Within the
past three years, this has become known to be completely false. And
the ear is an extension of the nervous system.

So what?

This paper set forth a "threshold of hearing":

H. Fletcher and W. A. Munson, "Loudness, its definition, measurement and
calculation," Journal of the Acoustical Society of America, vol. 5, pp.
82--108, Oct. 1933.

I think that the last paper investigating the same property of the ears I've
seen was dated some time in early 2002. It basically confirmed the same
results.

Yet delusional golden ear audiophiles such as yourself Morein, like to
disregard the idea that the human ear has known, finite sensitivity limits
that are actually quite high by modern technical standards.

Interestingly the way some people interpreted the F-M results was found to
be wildly optimistic by a number of scientists about 10 years ago, guys like
Zwicker and Fastl. Bottom line, in actual use the ear often disregards far
higher levels of sound than might be predicted by a naive reading of F-M.

Did Z-W know what they are talking about? A whole segment of the audio
industry based on "Perceptual Coding" is built on the scientific findings
they enlightened us about. The net of "Perceptual Coding" is that something
like 90% or more distortion (in the sense of loss of information) is
acceptable to the ear, if you pick the right 90% to distort.

Your statement above may be correct, or it may not, but there is no
good science to back it up.

There's tons of good science and empirical evidence that backs it up. If
you want to point to some minor point about the mammalian brain as proof
that none of the "soft sciences" are any good, that's up to you. But that
isn't exactly very good logic, is it?

All there is are isolated studies, flawed
or not, which may be used to incorrectly extrapolate the conclusion
that perfection at the level of a Halcro is irrelevant.

Consummate ignorance of the relevant science is noted.

Bad scientist alert STANDS!!!

You just indicted yourself as a bad scientist again, Morein. Thanks!

Obsess over the Halcro if you want to Morein, it's your money, your life.
Just don't try to use egregiously bad logic like this and your ignorance of
what's known about the sensitivity of the human ear to justify your
obsessions and mislead others.

Robert Morein wrote:

Because soft science relies so much on observation, without any overarching
theory to back it up, it is prone to error and reversal. For example, for a
hundred years, it was believed that neuronal replacement did not occur in
the adult mammalian brain. Within the past three years, this has become
known to be completely false. And the ear is an extension of the nervous
system.

Actually, people knew this over a decade ago - by witnessing such things as
people who had half of their brain missing and seeing it re-wire itself
into the non-dead portions.

Of course, like most things in life, widespread acceptance takes about
a decade or two to happen.

"Robert Morein" wrote in message
...

Oddly, although amplifier design has become mature, it does not appear to
me
that specification and testing has. The scientist in me says it all has to
reduce to numbers, yet it does not...

Actually it does, but we've simply failed to create devices that accurately
measure the attributes we (audiophiles) are interested in.

"jeffc" wrote in message
m

"Robert Morein" wrote in message
...

Oddly, although amplifier design has become mature, it does not
appear to me that specification and testing has. The scientist in me
says it all has to reduce to numbers, yet it does not...

Actually it does, but we've simply failed to create devices that
accurately measure the attributes we (audiophiles) are interested in.

Most such attributes being non-sonic. That's a problem with the soft science
of marketing, not the hard science of measurement of audio parameters.

"John Atkinson" wrote in message
om

(Audio Guy) wrote on r.a.h-e in message
...

Audio is a trivial application, they learn about power supply
design and amplification, which is pretty much all there is to
audio amplifiers, in their early years and then go on to much more
interesting and challenging concepts.

I see statements like this from time to time, yet I am not so sure
that audio design is "trivial."

One of the better online discussions of audio power amp design is posted at
http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm . Not trivial, but also
no longer any kind of big secret for people who do their homework.

There are not many other design fields
where an amplifier: has to provide up to 30dB of voltage gain; act as
a voltage source into a wide and arbitrary range of load impedances
and do so in an unconditionally stable manner; have a passband noise
contribution at least 90dB down from 1W into 8 ohms, no matter what
its voltage gain and ultimate power delivery; have distortion
components under all load conditions that are below the threshold of
hearing no matter what the program material is; and do all the above
over at least three-decade, ie, a 10-octave passband.

However, for $5 or $10 you can buy one-up, a chip that delivers very usable
amounts of power with very little or no audible distortion and just a few
added parts. However, turning this chip into a competitive product is still
a goodly amount of work.

Thoughts, gentlemen? I would suggest that designing, say, a typical RF
amplifier is, by comparison, "trivial" but, of course, I may just be
missing something :-)

Any project that involves designing and/or building and selling a
competitive product is challenging. Even the experts fail at it, every once
in a while.

In article ,
John Atkinson wrote:
(Audio Guy) wrote on r.a.h-e in message
...
Audio is a trivial application, they learn about power supply
design and amplification, which is pretty much all there is to
audio amplifiers, in their early years and then go on to much more
interesting and challenging concepts.

I see statements like this from time to time, yet I am not so sure
that audio design is "trivial." There are not many other design fields
where an amplifier: has to provide up to 30dB of voltage gain; act as a
voltage source into a wide and arbitrary range of load impedances
and do so in an unconditionally stable manner; have a passband noise

Non-trivial!


Thoughts, gentlemen? I would suggest that designing, say, a typical RF
amplifier is, by comparison, "trivial" but, of course, I may just be
missing something :-)


Indeed, so.


Joe

In article ,
(John Atkinson) writes:
(Audio Guy) wrote on r.a.h-e in message
...
Audio is a trivial application, they learn about power supply
design and amplification, which is pretty much all there is to
audio amplifiers, in their early years and then go on to much more
interesting and challenging concepts.

I see statements like this from time to time, yet I am not so sure
that audio design is "trivial." There are not many other design fields
where an amplifier: has to provide up to 30dB of voltage gain; act as a
voltage source into a wide and arbitrary range of load impedances
and do so in an unconditionally stable manner; have a passband noise
contribution at least 90dB down from 1W into 8 ohms, no matter what
its voltage gain and ultimate power delivery; have distortion
components under all load conditions that are below the threshold of
hearing no matter what the program material is; and do all the above
over at least three-decade, ie, a 10-octave passband.

Thoughts, gentlemen? I would suggest that designing, say, a typical RF
amplifier is, by comparison, "trivial" but, of course, I may just be
missing something :-)

Sure, take my words out of context.

You left out the previous part that qualifies them:

"Audiophiles don't realize that audio is an extremely small
part of electrical engineering and that very, very few schools even
teach courses in the subject. It isn't where the money is, nor is it
where the interest is for EE students. EEs like to make ICs or design
computers or work in motor control or design antennas or work in
telecom."

(Audio Guy) wrote in message
...
In article ,
(John Atkinson) writes:
(Audio Guy) wrote on r.a.h-e in message
...
Audio is a trivial application, they learn about power supply
design and amplification, which is pretty much all there is to
audio amplifiers, in their early years and then go on to much more
interesting and challenging concepts.

I see statements like this from time to time, yet I am not so sure
that audio design is "trivial." There are not many other design fields
where an amplifier: has to provide up to 30dB of voltage gain; act as a
voltage source into a wide and arbitrary range of load impedances
and do so in an unconditionally stable manner; have a passband noise
contribution at least 90dB down from 1W into 8 ohms, no matter what
its voltage gain and ultimate power delivery; have distortion
components under all load conditions that are below the threshold of
hearing no matter what the program material is; and do all the above
over at least three-decade, ie, a 10-octave passband.

Thoughts, gentlemen? I would suggest that designing, say, a typical RF
amplifier is, by comparison, "trivial" but, of course, I may just be
missing something :-)

Sure, take my words out of context.

You left out the previous part that qualifies them:
"Audiophiles don't realize that audio is an extremely small
part of electrical engineering and that very, very few schools even
teach courses in the subject. It isn't where the money is, nor is it
where the interest is for EE students. EEs like to make ICs or design
computers or work in motor control or design antennas or work in
telecom."

My apologies "audioguy" but I don't see how this paragraph changes the
meaning of the words you wrote about audio amplifier design. It doesn't
matter _ why_ electronic engineers feel audio is a "trivial" application,
only that they do, and that is what I was addressing. Note that I feel
that audio amplifier design is far from trivial. If you look at the
list of attributes I listed for an ideal audio amplifier, I can think of
almost none, of all the designs I have tested for Stereophile, that
achieve that level of performance.

John Atkinson
Editor, Stereophile

In article ,
(John Atkinson) writes:
(Audio Guy) wrote in message
...
In article ,
(John Atkinson) writes:
(Audio Guy) wrote on r.a.h-e in message
...
Audio is a trivial application, they learn about power supply
design and amplification, which is pretty much all there is to
audio amplifiers, in their early years and then go on to much more
interesting and challenging concepts.

I see statements like this from time to time, yet I am not so sure
that audio design is "trivial." There are not many other design fields
where an amplifier: has to provide up to 30dB of voltage gain; act as a
voltage source into a wide and arbitrary range of load impedances
and do so in an unconditionally stable manner; have a passband noise
contribution at least 90dB down from 1W into 8 ohms, no matter what
its voltage gain and ultimate power delivery; have distortion
components under all load conditions that are below the threshold of
hearing no matter what the program material is; and do all the above
over at least three-decade, ie, a 10-octave passband.

Thoughts, gentlemen? I would suggest that designing, say, a typical RF
amplifier is, by comparison, "trivial" but, of course, I may just be
missing something :-)

Sure, take my words out of context.

You left out the previous part that qualifies them:
"Audiophiles don't realize that audio is an extremely small
part of electrical engineering and that very, very few schools even
teach courses in the subject. It isn't where the money is, nor is it
where the interest is for EE students. EEs like to make ICs or design
computers or work in motor control or design antennas or work in
telecom."

My apologies "audioguy" but I don't see how this paragraph changes the
meaning of the words you wrote about audio amplifier design. It doesn't
matter _ why_ electronic engineers feel audio is a "trivial" application,
only that they do, and that is what I was addressing. Note that I feel
that audio amplifier design is far from trivial. If you look at the
list of attributes I listed for an ideal audio amplifier, I can think of
almost none, of all the designs I have tested for Stereophile, that
achieve that level of performance.

My point is that relative to other areas of EE, audio IS trivial. As
well as mundane and much less lucrative. I mean, come on, it just
doesn't compare to to something like designing ICs at the sub-micron
level, or optical transmission at a 10 Gbit/sec rate.

"Audio Guy" wrote in message
...
In article ,
(John Atkinson) writes:
(Audio Guy) wrote in message
...
In article ,
(John Atkinson) writes:
(Audio Guy) wrote on r.a.h-e in message
...
Audio is a trivial application, they learn about power supply
design and amplification, which is pretty much all there is to
audio amplifiers, in their early years and then go on to much more
interesting and challenging concepts.

I see statements like this from time to time, yet I am not so sure
that audio design is "trivial." There are not many other design
fields
where an amplifier: has to provide up to 30dB of voltage gain; act as
a
voltage source into a wide and arbitrary range of load impedances
and do so in an unconditionally stable manner; have a passband noise
contribution at least 90dB down from 1W into 8 ohms, no matter what
its voltage gain and ultimate power delivery; have distortion
components under all load conditions that are below the threshold of
hearing no matter what the program material is; and do all the above
over at least three-decade, ie, a 10-octave passband.

Thoughts, gentlemen? I would suggest that designing, say, a typical
RF
amplifier is, by comparison, "trivial" but, of course, I may just be
missing something :-)

Sure, take my words out of context.

You left out the previous part that qualifies them:
"Audiophiles don't realize that audio is an extremely small
part of electrical engineering and that very, very few schools even
teach courses in the subject. It isn't where the money is, nor is it
where the interest is for EE students. EEs like to make ICs or design
computers or work in motor control or design antennas or work in
telecom."

My apologies "audioguy" but I don't see how this paragraph changes the
meaning of the words you wrote about audio amplifier design. It doesn't
matter _ why_ electronic engineers feel audio is a "trivial"
application,
only that they do, and that is what I was addressing. Note that I feel
that audio amplifier design is far from trivial. If you look at the
list of attributes I listed for an ideal audio amplifier, I can think of
almost none, of all the designs I have tested for Stereophile, that
achieve that level of performance.

My point is that relative to other areas of EE, audio IS trivial. As
well as mundane and much less lucrative. I mean, come on, it just
doesn't compare to to something like designing ICs at the sub-micron
level, or optical transmission at a 10 Gbit/sec rate.

The latest EE Times has an article, "40 gHz and Beyond"

40 gHz!
Of all the centuries, this is the most unlike any other.
I've been told I look about 27 years old, yet I remember futzing with 6BA6's
and other thermionic devices.
When I was a child, I took a Coke-bottle tube out of the back of the TV and
held it in my hand, much to the consternation of my baby sitter.
Then I remember staring at a 2764 under a Nikon binocular microscope, and
marveled at the tiny array.
With a 27C512, you can't see the array!
Now IC lithography is going beyond the realm of visible light, into UV and
soft X-ray.

Hell, when I was a physics grad student, we used to believe that the
two-slit diffraction experiment was explained by Heisenberg's Uncertainty
Principle.
My dear qmech teacher, Sigurd Larsen, showed us that all "Hidden Variables
Theories" had inherent contradictions.
Turned out VonNeumann's proof was wrong!

Next up: Spintronics, on the Road to 100 gHz Computing!

"Audio Guy" wrote in message


My point is that relative to other areas of EE, audio IS trivial.

Well certainly, most legacy audio is very cut-and-dried. Other than making
high quality audio smaller/lighter/cheaper there isn't a lot of life left in
the old girl.

Audio could be more interesting and profitable, but so much of its real and
intangible capital has been squandered on snake oil. Audio has been largely
a derivative art for at least 20-30 years.

As well as mundane and much less lucrative.

Bingo, and one reason why I've never looked for a full-time job in audio
since I was 15. I have some friends who did with varied success, but many of
them either fell off the gravy train or are glad they are coming up on
retirement.

I mean, come on, it just
doesn't compare to to something like designing ICs at the sub-micron
level, or optical transmission at a 10 Gbit/sec rate.

There are a few ways to make the big bucks in audio and not soil oneself
with snake oil, but they are few and far between.

"John Atkinson" wrote in message
om...
(Audio Guy) wrote in message
...
In article ,
(John Atkinson) writes:
(Audio Guy) wrote on r.a.h-e in message
...
Audio is a trivial application, they learn about power supply
design and amplification, which is pretty much all there is to
audio amplifiers, in their early years and then go on to much more
interesting and challenging concepts.

I see statements like this from time to time, yet I am not so sure
that audio design is "trivial." There are not many other design fields
where an amplifier: has to provide up to 30dB of voltage gain; act as
a
voltage source into a wide and arbitrary range of load impedances
and do so in an unconditionally stable manner; have a passband noise
contribution at least 90dB down from 1W into 8 ohms, no matter what
its voltage gain and ultimate power delivery; have distortion
components under all load conditions that are below the threshold of
hearing no matter what the program material is; and do all the above
over at least three-decade, ie, a 10-octave passband.

Thoughts, gentlemen? I would suggest that designing, say, a typical RF
amplifier is, by comparison, "trivial" but, of course, I may just be
missing something :-)

Sure, take my words out of context.

You left out the previous part that qualifies them:
"Audiophiles don't realize that audio is an extremely small
part of electrical engineering and that very, very few schools even
teach courses in the subject. It isn't where the money is, nor is it
where the interest is for EE students. EEs like to make ICs or design
computers or work in motor control or design antennas or work in
telecom."

My apologies "audioguy" but I don't see how this paragraph changes the
meaning of the words you wrote about audio amplifier design. It doesn't
matter _ why_ electronic engineers feel audio is a "trivial" application,
only that they do, and that is what I was addressing. Note that I feel
that audio amplifier design is far from trivial. If you look at the
list of attributes I listed for an ideal audio amplifier, I can think of
almost none, of all the designs I have tested for Stereophile, that
achieve that level of performance.

John Atkinson
Editor, Stereophile

Many words slip into a discourse through inadvertent prejudice or
association, rather than intended meaning.
I think what Audioguy meant to say is that audio amplification is regarded
as trivial not because it is, but because it is bereft of the prestige that
comes with working in a field with actively advancing fundamentals.
For example, Kalman filtering is no less important now than in the early
1960's, yet only the specialists who actually insert these very canonical
algorithms into microcontrollers pay any attention.

Examples of recent, really remarkable advances which incorporate elements
of circuit theory are
high speed serial bus transducers, such as RAMBUS or USB
flash A/D
the neodymium lightwave amplfier
Sun's capacitive chip interface
and a little bit earlier: the gyrator

By contrast, designing and building an audio amplifier is somewhat like
cutting a diamond. It has become an almost timeless skill.

But I say to you, John, that nobody has come up with a reasonable set of
figures of merit. I still can't look at a set of your graphs and predict how
an amp will sound. That's a disconnect.

"Robert Morein" wrote in message


But I say to you, John, that nobody has come up with a reasonable set
of figures of merit.

Admittedly he hasn't dumbed his tech reports down enough to address the
typical nontechnical audiophile. However, these guys are generally snowed by
the subjectivist poetry and song-and-dance.

I still can't look at a set of your graphs and predict how an amp will
sound.

Confession of highly ability to make abstraction relate to the real world
noted.

John's charts and graphs could be used to sift the sonically transparent
amps from the others. For the ones that aren't transparent, a fairly
close-sounding model could be constructed from his reports. What more does
it take?

That's a disconnect.

Only for people who aren't up-to-date about science and audio and don't want
to bother to learn.

"Robert Morein" wrote in message
...
I think what Audioguy meant to say is that audio amplification is
regarded as trivial not because it is, but because it is bereft of the
prestige that comes with working in a field with actively advancing
fundamentals.

If so, then I don't disagree.

I say to you, John, that nobody has come up with a reasonable set of
figures of merit. I still can't look at a set of your graphs and
predict how an amp will sound. That's a disconnect.

Unfortunately yes. But two excellent papers by Earl Geddes and his wife,
presented at the recent AES Convention, show a way forward, by looking at
how the spuriae produced by the "bent" transfer function of a typical
amplifier can be examined using a masking model representing human
perception. Their provisional results show excellent correlation between
the metric for a given amplifier and the audibility of its spuriae.

This work is not at the stage where someone could plug measured results
into a spreadsheet and out pops a "good" "moderate" "bad" judgment, but
eventually something like that will be possible. Of course, a human
reviewer will still be needed to produce what Tom Nosuaine calls the
"audio poetry." :-)

John Atkinson
Editor, Stereophile

On 23 Oct 2003 14:13:21 -0700, (John
Atkinson) wrote:

I say to you, John, that nobody has come up with a reasonable set of
figures of merit. I still can't look at a set of your graphs and
predict how an amp will sound. That's a disconnect.

Unfortunately yes. But two excellent papers by Earl Geddes and his wife,
presented at the recent AES Convention, show a way forward, by looking at
how the spuriae produced by the "bent" transfer function of a typical
amplifier can be examined using a masking model representing human
perception. Their provisional results show excellent correlation between
the metric for a given amplifier and the audibility of its spuriae.

Damn...you're making my head hurt.

"John Atkinson" wrote in message
om...
"Robert Morein" wrote in message
...
I think what Audioguy meant to say is that audio amplification is
regarded as trivial not because it is, but because it is bereft of the
prestige that comes with working in a field with actively advancing
fundamentals.

If so, then I don't disagree.

I say to you, John, that nobody has come up with a reasonable set of
figures of merit. I still can't look at a set of your graphs and
predict how an amp will sound. That's a disconnect.

Unfortunately yes. But two excellent papers by Earl Geddes and his wife,
presented at the recent AES Convention, show a way forward, by looking at
how the spuriae produced by the "bent" transfer function of a typical
amplifier can be examined using a masking model representing human
perception. Their provisional results show excellent correlation between
the metric for a given amplifier and the audibility of its spuriae.

That sounds very promising.
There may yet be room for audio poetry, however.
If I may liken audio to a data stream with "features", the mere audibility
of a feature "discrepancy" may not correlate with the emotional importance
to the listener. The frequency of the "feature", and thus the "discrepancy",
also depends upon the musical material. Treble grain may be barely audible,
and comprise an insignificant portion of the power spectrum, yet be as
irritating as chalk on a blackboard. And loose, tubby bass may be considered
a benefit to some.

The most consistent thing about amplifiers I have noticed is my personal
need to match the amplifier to the speaker. Bright amps work well with soft
domes, while softer, ie., MOSFET amps, work better for me with hard domes.
The right wine for the meal.

But I have had in the back of my mind using a Hafler type bridge to get the
difference signal and look at the spectra.
I was largely dissuaded by the thought that this was a solution that nobody
wanted. as many members of this group might state.

This work is not at the stage where someone could plug measured results
into a spreadsheet and out pops a "good" "moderate" "bad" judgment, but
eventually something like that will be possible. Of course, a human
reviewer will still be needed to produce what Tom Nosuaine calls the
"audio poetry." :-)

John Atkinson
Editor, Stereophile

(John Atkinson) wrote:


"Robert Morein" wrote in message
...
I think what Audioguy meant to say is that audio amplification is
regarded as trivial not because it is, but because it is bereft of the
prestige that comes with working in a field with actively advancing
fundamentals.

If so, then I don't disagree.

I say to you, John, that nobody has come up with a reasonable set of
figures of merit. I still can't look at a set of your graphs and
predict how an amp will sound. That's a disconnect.

Unfortunately yes. But two excellent papers by Earl Geddes and his wife,
presented at the recent AES Convention, show a way forward, by looking at
how the spuriae produced by the "bent" transfer function of a typical
amplifier can be examined using a masking model representing human
perception. Their provisional results show excellent correlation between
the metric for a given amplifier and the audibility of its spuriae.

OK I heard that paper. So how does that show that any ampliifer on the market
currently available or on your RCL (congratulations on learning how to count)
is "better sounding" than any other with regard to the GedLee Metric?


This work is not at the stage where someone could plug measured results
into a spreadsheet and out pops a "good" "moderate" "bad" judgment, but
eventually something like that will be possible. Of course, a human
reviewer will still be needed to produce what Tom Nosuaine calls the
"audio poetry." :-)

So how are your bais-controlled validation listening tests going? I'm comforted
that you'll be there for the "poetry" part. Wouldn't it be horrible for
enthusiasts to learn from a spec-sheet "number" that their Yamaha integrated
amplifier sounds exactly like a 12K high-end set of mono-blocks ?

"John Atkinson" wrote in message
om

Unfortunately yes. But two excellent papers by Earl Geddes and his
wife, presented at the recent AES Convention, show a way forward, by
looking at how the spuriae produced by the "bent" transfer function
of a typical amplifier can be examined using a masking model
representing human perception. Their provisional results show
excellent correlation between the metric for a given amplifier and
the audibility of its spuriae.

Geddes' specialty is loudspeaker design and evaluation. Therefore he's
concerned with relatively high levels of distortion by the standards of
modern audio power amplifiers.

"John Atkinson" wrote in message
om...
(Audio Guy) wrote in message
...
In article ,
(John Atkinson) writes:
(Audio Guy) wrote on r.a.h-e in message
...
Audio is a trivial application, they learn about power supply
design and amplification, which is pretty much all there is to
audio amplifiers, in their early years and then go on to much
more
interesting and challenging concepts.

I see statements like this from time to time, yet I am not so
sure
that audio design is "trivial." There are not many other design
fields
where an amplifier: has to provide up to 30dB of voltage gain;
act as a
voltage source into a wide and arbitrary range of load
impedances
and do so in an unconditionally stable manner; have a passband
noise
contribution at least 90dB down from 1W into 8 ohms, no matter
what
its voltage gain and ultimate power delivery; have distortion
components under all load conditions that are below the
threshold of
hearing no matter what the program material is; and do all the
above
over at least three-decade, ie, a 10-octave passband.

Thoughts, gentlemen? I would suggest that designing, say, a
typical RF
amplifier is, by comparison, "trivial" but, of course, I may
just be
missing something :-)

Sure, take my words out of context.

You left out the previous part that qualifies them:
"Audiophiles don't realize that audio is an extremely small
part of electrical engineering and that very, very few schools
even
teach courses in the subject. It isn't where the money is, nor is
it
where the interest is for EE students. EEs like to make ICs or
design
computers or work in motor control or design antennas or work in
telecom."

My apologies "audioguy" but I don't see how this paragraph changes
the
meaning of the words you wrote about audio amplifier design. It
doesn't
matter _ why_ electronic engineers feel audio is a "trivial"
application,
only that they do, and that is what I was addressing. Note that I
feel
that audio amplifier design is far from trivial. If you look at the
list of attributes I listed for an ideal audio amplifier, I can
think of
almost none, of all the designs I have tested for Stereophile, that
achieve that level of performance.

John Atkinson
Editor, Stereophile

Certainly more than half of the requirements you placed on your
amplifier design detailed in your first reply to this header disappear
if the designer knows the characteristics of the source and load ahead
of time. It is for this reason that I forever wonder at how few
speakers come with their own amplifier--at least in the home hi-fi
industry.

Norm Strong

"normanstrong" wrote in message
news:lmVlb.3955$mZ5.23026@attbi_s54...
Certainly more than half of the requirements you placed on your
amplifier design detailed in your first reply to this header disappear
if the designer knows the characteristics of the source and load ahead
of time. It is for this reason that I forever wonder at how few
speakers come with their own amplifier--at least in the home hi-fi
industry.

Hi Norm, I believe that the apparent restriction of customer choice that
this represents is a major impediment to successful marketing of an active
speaker. Even if the customer buys exactly the amplifier that the speaker
designer feels works best with his loudspeaker, and would therefore be
the one that could be supplied in an integrated package, it appears to be
important to customers to have the widest amplifier choice available.

Probably only Meridian has made much headway in the audiophile market
selling integrated loudspeaker/amplifier packages.

John Atkinson
Editor, Stereophile

PS: Your letter on the purported advantages of hi-rez audio media
appears in the ne (November) issue of Stereophile.

(John Atkinson) wrote:

"normanstrong" wrote in message
news:lmVlb.3955$mZ5.23026@attbi_s54...
Certainly more than half of the requirements you placed on your
amplifier design detailed in your first reply to this header disappear
if the designer knows the characteristics of the source and load ahead
of time. It is for this reason that I forever wonder at how few
speakers come with their own amplifier--at least in the home hi-fi
industry.

Hi Norm, I believe that the apparent restriction of customer choice that
this represents is a major impediment to successful marketing of an active
speaker. Even if the customer buys exactly the amplifier that the speaker
designer feels works best with his loudspeaker, and would therefore be
the one that could be supplied in an integrated package, it appears to be
important to customers to have the widest amplifier choice available.

I have a different take on this. An average "high-end" enthusiast may already
have a stack of amplifiers or at least a single expensive 2-channel amplifier
at his/her disposal. Anybody else has purchased a modrn receiver with 5
amplifier channels already built into the chassis.

People just don't see the reason to buy power again. The neo-phytes don't get
it. The high-enders never WILL get it; it's too apparent. So the best way to
better sounding loudspeakers will remain foreign to the market UNTIL the
current generation of computer users (who expect their speakers to be powered)
are in the loop.


Probably only Meridian has made much headway in the audiophile market
selling integrated loudspeaker/amplifier packages.

John Atkinson
Editor, Stereophile

PS: Your letter on the purported advantages of hi-rez audio media
appears in the ne (November) issue of Stereophile.

That take is likely a fair statement of the current market. Paradigm stopped
offering their Active series which was one of the better performing speaker
systems ever to reach the market. Yet few would buy them.

"normanstrong" wrote:

....snips.....


"John Atkinson" wrote in message

I see statements like this from time to time, yet I am not so
sure
that audio design is "trivial." There are not many other design
fields
where an amplifier: has to provide up to 30dB of voltage gain;
act as a
voltage source into a wide and arbitrary range of load
impedances
and do so in an unconditionally stable manner; have a passband
noise
contribution at least 90dB down from 1W into 8 ohms, no matter
what
its voltage gain and ultimate power delivery; have distortion
components under all load conditions that are below the
threshold of
hearing no matter what the program material is; and do all the
above
over at least three-decade, ie, a 10-octave passband.

Thoughts, gentlemen? I would suggest that designing, say, a
typical RF
amplifier is, by comparison, "trivial" but, of course, I may
just be
missing something :-)

I don't know if Mr Atkinson is missing anything. But I wonder exactly how it is
that with the exception of high-ouput impedance amplifiers that exactly NO ONE
hasd ever shown in a reasonably well bias-controlled experiment (including Mr
Atkinson's; indeed his personally conducted large trial experiments showed that
even with high-output impedance amplifiers there was no sonic difference) that
modern amplfiers have any sound of their own.



Sure, take my words out of context.

You left out the previous part that qualifies them:
"Audiophiles don't realize that audio is an extremely small
part of electrical engineering and that very, very few schools
even
teach courses in the subject. It isn't where the money is, nor is
it
where the interest is for EE students. EEs like to make ICs or
design
computers or work in motor control or design antennas or work in
telecom."

My apologies "audioguy" but I don't see how this paragraph changes
the
meaning of the words you wrote about audio amplifier design. It
doesn't
matter _ why_ electronic engineers feel audio is a "trivial"
application,
only that they do, and that is what I was addressing. Note that I
feel
that audio amplifier design is far from trivial. If you look at the
list of attributes I listed for an ideal audio amplifier, I can
think of
almost none, of all the designs I have tested for Stereophile, that
achieve that level of performance.

John Atkinson
Editor, Stereophile


OK; which ones that you recommend have been shown to sound different with a
bias controlled listening test?


Certainly more than half of the requirements you placed on your
amplifier design detailed in your first reply to this header disappear
if the designer knows the characteristics of the source and load ahead
of time. It is for this reason that I forever wonder at how few
speakers come with their own amplifier--at least in the home hi-fi
industry.

Norm Strong

Actually the best-performing speaker I've tested DO come with their own
amplifiers and application specific EQ.

I wonder what Mr Atkinson's experience may be in this regard.

"Nousaine" wrote in message
...
"normanstrong" wrote:

...snips.....


"John Atkinson" wrote in message

I see statements like this from time to time, yet I am not so
sure
that audio design is "trivial." There are not many other design
fields
where an amplifier: has to provide up to 30dB of voltage gain;
act as a
voltage source into a wide and arbitrary range of load
impedances
and do so in an unconditionally stable manner; have a passband
noise
contribution at least 90dB down from 1W into 8 ohms, no matter
what
its voltage gain and ultimate power delivery; have distortion
components under all load conditions that are below the
threshold of
hearing no matter what the program material is; and do all the
above
over at least three-decade, ie, a 10-octave passband.

Thoughts, gentlemen? I would suggest that designing, say, a
typical RF
amplifier is, by comparison, "trivial" but, of course, I may
just be
missing something :-)

I don't know if Mr Atkinson is missing anything. But I wonder exactly how
it is
that with the exception of high-ouput impedance amplifiers that exactly NO
ONE
hasd ever shown in a reasonably well bias-controlled experiment (including
Mr
Atkinson's; indeed his personally conducted large trial experiments showed
that
even with high-output impedance amplifiers there was no sonic difference)
that
modern amplfiers have any sound of their own.



Sure, take my words out of context.

You left out the previous part that qualifies them:
"Audiophiles don't realize that audio is an extremely small
part of electrical engineering and that very, very few schools
even
teach courses in the subject. It isn't where the money is, nor is
it
where the interest is for EE students. EEs like to make ICs or
design
computers or work in motor control or design antennas or work in
telecom."

My apologies "audioguy" but I don't see how this paragraph changes
the
meaning of the words you wrote about audio amplifier design. It
doesn't
matter _ why_ electronic engineers feel audio is a "trivial"
application,
only that they do, and that is what I was addressing. Note that I
feel
that audio amplifier design is far from trivial. If you look at the
list of attributes I listed for an ideal audio amplifier, I can
think of
almost none, of all the designs I have tested for Stereophile, that
achieve that level of performance.

John Atkinson
Editor, Stereophile


OK; which ones that you recommend have been shown to sound different with
a
bias controlled listening test?


Certainly more than half of the requirements you placed on your
amplifier design detailed in your first reply to this header disappear
if the designer knows the characteristics of the source and load ahead
of time. It is for this reason that I forever wonder at how few
speakers come with their own amplifier--at least in the home hi-fi
industry.

Norm Strong

Actually the best-performing speaker I've tested DO come with their own
amplifiers and application specific EQ.

I wonder what Mr Atkinson's experience may be in this regard.

What speaker do you refer to?

(John Atkinson) wrote in message . com...
(Audio Guy) wrote on r.a.h-e in message
...
Audio is a trivial application, they learn about power supply
design and amplification, which is pretty much all there is to
audio amplifiers, in their early years and then go on to much more
interesting and challenging concepts.

I see statements like this from time to time, yet I am not so sure
that audio design is "trivial." There are not many other design fields
where an amplifier: has to provide up to 30dB of voltage gain; act as a
voltage source into a wide and arbitrary range of load impedances
and do so in an unconditionally stable manner; have a passband noise
contribution at least 90dB down from 1W into 8 ohms, no matter what
its voltage gain and ultimate power delivery; have distortion
components under all load conditions that are below the threshold of
hearing no matter what the program material is; and do all the above
over at least three-decade, ie, a 10-octave passband.

Thoughts, gentlemen? I would suggest that designing, say, a typical RF
amplifier is, by comparison, "trivial" but, of course, I may just be
missing something :-)


I agree that designing a state of the art audio amplifier is not
trivial, for the reasons you stated above.

Designing an RF amplifier is more challenging than designing a audio
power amplifier. For example, a typical CATV RF amplifier would be 40
MHz - 1 GHz, and have 18 to 33 dB gain. Its' basic layout is very
similar to an audio power amplifier. The CATV amp will have: a couple
push-pull stages of bipolar transistors, a requirement for flat
response, and for very low second and third order distortion.

I don't think you would be suprised to hear that getting an amplifier
flat out to 1 GHz is more difficult than get an amplifier flat out to
20 kHz. In addition, CATV amplifiers are cascaded, (sometimes up to 50
amps in cascade). The sum total of the flatness of the cascade is
typically better than 2 dB. That works out to 1/25 dB flatness per
amplifier.

The visability of noise and distortion on video, are roughly the same
as the audibilty of noise and distortion for audio. But because noise
and distortion add in a cascade, the distortion performance of each
individual CATV amplifier must be better than that required for audio
amplifiers.

Stability of amplifiers that work up to a 1 GHz is more difficult to
achieve than for amplifiers that work only up to 20 KHz. Although RF
amps typically work into a 75 Ohm loads, they must be designed to be
stable into any load from a short circuit, to a complete open.

The list differences could go on, but if you pick up a textbook on
audio design, and one on RF design, you will see that the RF design is
or complex.

Bob Stanton

"Bob-Stanton" wrote in message
om...
(John Atkinson) wrote in message
. com...
(Audio Guy) wrote on r.a.h-e in message
...
Audio is a trivial application, they learn about power supply
design and amplification, which is pretty much all there is to
audio amplifiers, in their early years and then go on to much more
interesting and challenging concepts.

I see statements like this from time to time, yet I am not so sure
that audio design is "trivial." There are not many other design fields
where an amplifier: has to provide up to 30dB of voltage gain; act as a
voltage source into a wide and arbitrary range of load impedances
and do so in an unconditionally stable manner; have a passband noise
contribution at least 90dB down from 1W into 8 ohms, no matter what
its voltage gain and ultimate power delivery; have distortion
components under all load conditions that are below the threshold of
hearing no matter what the program material is; and do all the above
over at least three-decade, ie, a 10-octave passband.

Thoughts, gentlemen? I would suggest that designing, say, a typical RF
amplifier is, by comparison, "trivial" but, of course, I may just be
missing something :-)


I agree that designing a state of the art audio amplifier is not
trivial, for the reasons you stated above.

Designing an RF amplifier is more challenging than designing a audio
power amplifier. For example, a typical CATV RF amplifier would be 40
MHz - 1 GHz, and have 18 to 33 dB gain. Its' basic layout is very
similar to an audio power amplifier. The CATV amp will have: a couple
push-pull stages of bipolar transistors, a requirement for flat
response, and for very low second and third order distortion.

[snip]
I did some Google reading, and it appears that commercial CATV amplifier is
a very demanding case. However, design of a CATV amplifier has different
problems; it is not simply the case that it is more complex.
Were it not for the cascade requirement, one could make the case that the
audio amplifier is a greater challenge. In fact, I am not aware of any audio
amplifier ever produced that could meet the "analogous" CATV specs.
For an uncascaded RF amplifier, some characteristics of the problem are
relaxed compared to the audio amp.
Since the RF amplifier is not baseband, distortion in the signal does not
appear directly in the modulated signal. Even with simple class AB designs,
very high power output can be obtained with a small component count, since
the distortion products are broadband and not concentrated around the
carrier. By contrast, since the audio amplifier is a baseband instrument,
the distortion products are largely concentrated in the passband.

CATV amplifiers operate into a controlled impedance. The requirement that
they must survive a short is not equivalent to the challenge which audio
amplifiers encounter, where the load is an electrodynamic mechanical system
that provides a load to the system which is a function of frequency, and,
when accounting for nonlinear effects, amplitude and mechanical state.

If an audio amplifier were required to meet a spec with 50 of them in
cascade, we would see much more of the Halcro level of engineering.

"Robert Morein" wrote in message


I did some Google reading, and it appears that commercial CATV amplifier is
a very demanding case. However, design of a CATV amplifier has different
problems; it is not simply the case that it is more complex.
Were it not for the cascade requirement, one could make the case that the
audio amplifier is a greater challenge. In fact, I am not aware of any audio
amplifier ever produced that could meet the "analogous" CATV specs.
For an uncascaded RF amplifier, some characteristics of the problem are
relaxed compared to the audio amp.

That's right, they have a different set of problems. There is a hidden
complexity in RF circuits. A componet at audio frequencies is simple
single impedance. At RF a resistor is a network of three componets
(resistance, lead inductance, body capacitance). All these

(Bob-Stanton) wrote in message om...
"Robert Morein" wrote in message


I did some Google reading, and it appears that commercial CATV amplifier is
a very demanding case. However, design of a CATV amplifier has different
problems; it is not simply the case that it is more complex.
Were it not for the cascade requirement, one could make the case that the
audio amplifier is a greater challenge. In fact, I am not aware of any audio
amplifier ever produced that could meet the "analogous" CATV specs.
For an uncascaded RF amplifier, some characteristics of the problem are
relaxed compared to the audio amp.

That's right, they have a different set of problems. There is a hidden
complexity in RF circuits. A componet at audio frequencies is simple
single impedance. At RF a resistor is a network of three componets
(resistance, lead inductance, body capacitance). All these....

parasitic reactances make the resistor a small, two terminal network,
whos impedance changes with frequency.

A 100 pf capacitor, with 0.5 inch leads, becomes an inductor above 200
MHz.

Fields couple RF stages together and give strange unpredictable
results. There is more complexity in an RF design than is apparent
from the schemetic.

Bob Stanton

(Bob-Stanton) said:

A 100 pf capacitor, with 0.5 inch leads, becomes an inductor above 200
MHz.

Fields couple RF stages together and give strange unpredictable
results. There is more complexity in an RF design than is apparent
from the schemetic.

In my opinion, the same goes for audio.
RFI is a much bigger problem than most people think.
Have you ever tried to keep an oscilloscope's probe in the air in
mid-city?
I've seen audio amplifiers producing distortions and even Dc shifts
due to a cellular phone in the vicinity.
that's easily measured, but not so easily cured.

--
Sander deWaal
Vacuum Audio Consultancy

"Robert Morein" wrote in message

CATV amplifiers operate into a controlled impedance. The requirement that
they must survive a short is not equivalent to the challenge which audio
amplifiers encounter, where the load is an electrodynamic mechanical system
that provides a load to the system which is a function of frequency, and,
when accounting for nonlinear effects, amplitude and mechanical state.


Yes, a speaker systems provide a load that is a complex impedance, and
one that can change slightly due to nonlinear effects.

The RF amplifer can see a greater load variation. It must of stable
with an accidental low impedance (short) located anywhere along the
transmission line. RF amps are typically tested by sliding a low
impedance (about 1 Ohm) along a sloted line, on the output of the
amplifier. Here is the impedance variation the RF amplifier sees as
the resistor is moved along the sloted line:

Wavelengths Impedance

0.025 1.0 +j 12
0.050 1.1 +j 25
0.075 1.2 +j 38
0.100 1.5 +j 55
0.125 2.0 +j 75
0.150 2.9 +j 103
0.175 4.9 +j 147
0.200 10.5 +j 230
0.225 40.6 +j 470
0.250 5625.0 +j 0.00
0.275 40.6 -j 470
0.300 10.5 -j 230
0.325 4.9 -j 147
0.350 2.9 -j 103
0.375 2.0 -j 75
0.400 1.5 -j 55
0.425 1.2 -j 38
0.450 1.1 -j 25
0.475 1.0 -j 12
0.500 1.0 -j 0.00


This variation of impedance (going from short, to inductive, to open,
to capacitive) is much greater than an audio amplifier would ever see
from and electrodynamic mechanical nonlinear system.

Bob Stanton

"Robert Morein" wrote in message

Since the RF amplifier is not baseband, distortion in the signal does not
appear directly in the modulated signal.

Interference beats do appear on the video (on the TV screen). If one
takes a signal generator and inserts an interfering signal (sinewave)
on an RF line, it will cause wavey lines to appear on video.

The the threshold of visibility is -60 dB. In audio terms, that would
be 0.01% distortion. The threshold of interference visibility for
video, seems to be about the same as the threshold of distortion
audibility for audio.


Even with simple class AB designs,
very high power output can be obtained with a small component count, since
the distortion products are broadband and not concentrated around the
carrier. By contrast, since the audio amplifier is a baseband instrument,
the distortion products are largely concentrated in the passband.


Audio amplifiers and CATV amplifiers are both broadband products. They
both face the same distortion challenges. A RF amplifier tested with
a signal of two sinewaves, say at 400 MHz and 500 MHz, will generate
exactly the same kind of second order and third order products as an
audio amplifier with a signal of 400 Hz and 500 Hz.

For example, the audio amplifier will generate sum and difference
distortion products of 100 Hz and 900 Hz. The RF amplifier will
gernerate sum and difference products of 100 MHz and 900 MHz. (And so
on, for all the other second and third order products.)

Bob Stanton

"Bob-Stanton" wrote in message
om

The threshold of visibility is -60 dB. In audio terms, that would
be 0.01% distortion.

-60 dB is 0.1%

The threshold of interference visibility for
video, seems to be about the same as the threshold of distortion
audibility for audio.

Yes, even under the most ideal conditions, audibility of nonlinear
distortion seems to go away someplace around or below 0.1%. Under non-ideal
conditions, several percent nonlinear distortion can be missed.

"Arny Krueger" wrote in message ...
"Bob-Stanton" wrote in message
om

The threshold of visibility is -60 dB. In audio terms, that would
be 0.01% distortion.

-60 dB is 0.1%


Yes, I agree. My mistake

-20 dB = 10% distortion
-40 dB = 1% distortion
-60 dB = 0.1% distortion

It just proves the old saying:
"There are three kinds of people in this world. Those who can do math
and those who can't." :-)

Bob Stanton