Push-Pull vs. Single Ended Amplifiers

[Peter J. Ullman]

For the last few years much publicity has been given to the supposed
advantages of single-ended vs. push-pull output stages.

Are there any circuit specialists in this group who could perhaps list
side-by-side the theoretical pros and cons of both of these design
approaches?

[J. Gordon Rankin]

Peter,

As there are days in the month (Year), there are differences in the
two technologies. As an engineer and a musician, I grew up with PP
technology. Until the early 90's when I made my first SE amplifier.
That was it, all my Mullard EL-34's and Tung-Sol 6550 were sold for
the alcaimed triode.

It is just more musical.

You will see a lot of publicity for PP because of the power and
speakers needng that power. But really there are a lot more SE
companions than you can count.

Try it you'll never leave it.

Gordon Rankin
Wavelength Audio

[Scott Frankland]

J. Gordon Rankin wrote:

> As there are days in the month (Year), there are differences in the
> two technologies. As an engineer and a musician, I grew up with PP
> technology. Until the early 90's when I made my first SE amplifier.
> That was it, all my Mullard EL-34's and Tung-Sol 6550 were sold for
> the alcaimed triode.

> It is just more musical.

Thank you, Gordon, for your guru-like opinion on this matter, but Mr.
Ullman did not ask for a testamonial, he asked for the technical
differences--

"Are there any circuit specialists in this group who could perhaps
list side-by-side the theoretical pros and cons of both of these
design approaches?"

Let me start by listing the SE problems. If necessary, I can follow up
with a critique of push-pull operation (or perhaps Mr. Rankin will
save me the trouble!). In a previous thread I proposed to explain why
I thought SE amps lacked transparency. This seems like a good
opportunity to address both issues simultaneously.

There are four fundamental impediments to transparency in the
single-ended tube amplifier (SETA):

I. The first problem: damping. SETA's with low feedback factors will
be poorly equipped to damp loudspeaker resonance's due to excessive
amplifier output impedance. This impedance is effectively a resistance
inserted between the amplifier and the loudspeaker--not unlike that
produced by a very thin pair of speaker cables. To a rough
approximation it is the internal impedance of the output tubes as
reduced by the turns ratio of the output transformer and by negative
feedback, if such is present.

Being effectively a buffer between the amplifier and the loudspeaker,
this output resistance reduces damping. The loudspeaker then uses this
resistance as a load for back-EMF. The most common result is a
disjointed time coherence between bass fundamentals and their upper
harmonics. This can affect the listener's sense of pace, timing, and
rhythm; an important aspect of musical communication.

This problem can be minimized by using triode output tubes and by
selecting a loudspeaker with high inherent damping. The relationship
between output resistance and frequency response is a related problem,
and is discussed below as a separate item.

II. If the load impedance varies with frequency (as it does in most
loudspeakers), the current supplied to the load will also vary due to
Ohm's law, I = V/R. In place of R we must substitute Z in order to
account for inductive and capacitive reactance, which being
essentially a resistance that varies with frequency. In addition, the
reactance shifts the phase of the voltage waveform relative to the
current waveform.

The first thing to realize is that the amplifier and loudspeaker are
effectively one. No loudspeaker can deliver a flat frequency response
if the signal voltage appearing across the amplifier terminals is
subject to variations in load impedance with frequency. The ability
of an amplifier to ignore such variations is sometimes called a
constant-voltage characteristic, although this appellation is somewhat
optimistic in that, to be truly constant, the amplifier's output
resistance would have to be zero.

Since loudspeaker impedance in general varies with frequency, the
presence of output resistance in the current path from amp to speaker
will tend to magnify those variations. The voltage variation that
appears across the amplifier terminals is not then the same that
appears across the speaker terminals, because it will be modified by
the reactive combinations that form between amp and speaker at various
frequencies. The speaker input voltage then ceases to be a simple
function of signal voltage, and instead becomes a complex function of
RC, RL, and RLC.

Thus, excessive output resistance produces the same effective result
as inserting an equalizer ahead of a low output resistance amplifier,
and adjusting the controls to follow the same non-linear frequency
response as for a high output resistance amplifier!

The effect of a changing load impedance can be seen in frequency
response graphs for amplifiers driving simulated reactive loads (for
which, see *Stereophile* and/or *Audio*). For amplifiers with high
output resistance, such loads always impair the uniformity of response
as compared to a purely resistive load. What's worse is that the
response will be different for each type of loudspeaker! This problem
can be minimized, however, by selecting a loudspeaker with a
relatively flat impedance curve.

Indeed, many of the better speakers are impedance compensated, in
order to obtain a more accurate filter function at the various
crossover points, and to present a more uniform load to the driving
amplifier. Alternatively, a manufacturer of SETA's might collaborate
with a speaker manufacturer to tailor a given loudspeaker to a
particular amplifier. This is the "system's approach" to audio design.
In this case, however, the end-user sacrifices interchangeability.

III. This brings us to the most fundamental problem with triode
SETA's: the concave-upward curvature of the triode plate
characteristics near the bottom of the load-line (i.e., for negative
signal swings). This curvature indicates increased non-linear plate
conductance, as well as increased internal impedance. Thus, harmonic
distortion increases on the negative swing, and damping is reduced on
the negative swing.

The resulting harmonic distortion will be predominantly even-order
because it is asymmetrical, i.e., it is confined to one or the other
half-cycles (the negative cycle in this case). Prior to clipping, the
majority of this distortion will be low-order in triodes, i.e., 2nd,
3rd, and 4th. Nonetheless, since distortion in audio amplifiers occurs
mainly at bass frequencies (where most of the power is generated), the
related distortion products will appear in the midrange!

For example, distortion that occurs with a 75Hz bass fundamental will
generate a 2nd harmonic at 150Hz; a 3rd harmonic at 225Hz; and a 4th
harmonic at 300Hz. As the level increases, the order of distortion
increases, causing products to appear higher up in the midrange.

Most tube amps suffer from excessive low bass distortion due to
transformer saturation, to inadequate primary inductance, or to phase
shift due to roll-off. This is one reason why tube amplifiers with
excessive distortion in the bass range often sound warmer or fuller in
the lower midrange, and more "transparent" in the upper midrange.

This is a euphonic effect that many listeners find pleasing. If the
resulting distortion becomes too strong, however, it can turn
perceived transparency into strident glassiness, and warmth into a
thick, muffling veil of sonic wool. Reducing bass distortion is thus a
critical factor for midrange transparency and neutrality.

A partial solution to the problem of negative-cycle curvature in
SETA's is the use of high-efficiency speakers. This allows the SETA to
operate more nearly within the bounds of it's "first watt," where
excursion is limited and distortion is low. So now we are faced with a
three-fold restriction upon the use of SETA's, i.e., speakers that are
(1) highly damped; (2) impedance compensated; and (3) highly
efficient.

IV. Which brings us to the final SETA-related problem: excessive
transformer-related distortion. The root of the problem is the
steady-state magnetic flux that arises in the output transformer due
to the large idle current required to obtain the necessary class-A
operation. The idle current in a SETA flows through the primary coil
of the output transformer, generating steady-state flux lines.

Of course, iron can readily support only a limited number of flux
lines--typically between 10-15 thousand lines per square centimeter,
depending upon the material. As the audio signal swings positive, the
current through the coil increases, thereby increasing the flux
density in the core. Above the knee of the B-H curve the iron quickly
becomes saturated. At this point, the positive-going output cycle is
COMPRESSED.

When the positive cycle is compressed and the negative cycle enters
curvature--as occurs near clipping--the distortion will be SYMMETRICAL
and therefore odd-order, just as in push-pull operation (although not
quite so strident, due to the more rounded negative cycle). At
clipping, the numeric order of the distortion products increases
(i.e., converges to a higher order). It is essential, therefore, to
minimize transformer saturation in single-ended amplifiers.

In order to minimize distortion due to magnetic flux saturation, the
core is often gapped. A well-chosen gap has a number of desirable
properties: (1) it reduces the DC magnetization; (2) it reduces the
tendency of primary inductance to vary with current flow; and (3) it
acts as a relief valve for excess magnetic pressure. Flux in the core
that increases beyond what the iron can readily support is then
shunted into the air gap. Since air cannot be saturated by flux,
high-order distortion due to core saturation is minimized.

In spite of the gap, the DC flowing through the primary can still
reduce primary inductance, causing low frequency problems. The
inductance can of course be restored by increasing the turns, but then
the core size must be increased as well. The two elements must remain
proportional because the coil acts to GENERATE flux lines while the
core acts to CONCENTRATE the flux lines. There is thus little point in
burying a tiny coil inside a whopping slab of iron.

Increasing the turns, however, will cause HIGH FREQUENCY LOSSES due to
the increased leakage inductance and interwinding capacitance. It
then becomes difficult to provide global feedback at high frequencies
in a SETA. It is equally problematical to provide feedback in a LOCAL
loop, because doing so reduces the input sensitivity of the output
stage. The AC swing of the driver stage must then be increased. In the
case of triode output tubes with local feedback, the driver swing
becomes difficult to obtain without interstage transformers or by
means of a bootstrapped driver, boosted by positive feedback (as in
the McIntosh).

On the face of it, there doesn't seem to be any readily available
solution to this final problem. Certainly, not for amps in the
commonly available power classes.

V. It has been shown that BOTH frequency extremes suffer greatly in
the SETA when compared to the push-pull amp. If the goal is to drive
conventional loudspeakers with a high degree of control and
faithfulness to the input signal, then there is only one solution that
I can see to the four problems outlined above.

The solution is three-fold: (1) ELIMINATE the output transformer; (2)
EXTEND the bandwidth into the MHz region (inasmuch as possible); and
(3) apply negative LOOP FEEDBACK. This may be accomplished by means of
multiple, paralleled output tubes. For example, given a feedback
factor of 20db, sixteen parallel 6550's (converted to triode to reduce
output impedance) will give a nominal damping factor of ten (as a
rule, a damping factor of ten is considered to be just acceptable).

The problem from a design standpoint will be in driving the input
capacitance of the sixteen parallel triodes. The solution to this
problem is readily obtained, but is not very economical. Persons
interested in doing so might wish to look up my patent (US patent
#5,498,996) entitled: "High Power, High Fidelity Tube Amplifier"
(subject to license for commercial applications).

Of course, given that my analysis is correct, the end result will be
an amp that sounds more like a good push-pull amp than a typical SETA.
Nonetheless, if desired, the SETA characteristics can be restored
simply by cutting back on the feedback. This causes the first three
problems, discussed above, to reappear. Presumably this will then
restore the euphonic elements of the single-ended sound. In any case,
the frequency extremes will be much improved by the absence of the
output transformer.

Let's face it, gentlemen (and you lurking ladies, too), a truly
transparent amplifier will not add ANY audible harmonics to the output
(nor anything else not related to the input). As we used to say in the
60's, feel free to roll your own--just remember, whatever you smoke is
subject to color your blue skies gray. Of course, the opposite is
sometimes true, as well. But in this case we are not talking about
transparency so much as about euphony. And that is an entirely
different subject altogether.

SELECTED REFERENCES

W.H. Anderson, "Measurement of Amplifier Internal Impedance,"
Audio, Sep 1958.

Martin Colloms, High Performance Loudspeakers, Pentech Press,
London, second edition, 1980.

N.H. Crowhurst, "What's All This About Damping?," Audio,
Sep 1958.

T.S. Gray, Applied Electronics, John Wiley & Sons, Inc., NY,
2nd ed., 1954.

Jeffrey H. Johnson, "Power Amplifiers and the Loudspeaker
Load," Audio, Aug 1977.

G.S.C. Lucas, "Distortion in Valve
Characteristics," Exp. Wireless, Nov 1931, pp. 595-598.

R.O. Rowlands, "Harmonic Distortion and Negative Feedback," Wireless
Eng., Jun 1953, pp. 133-135.

J.D. Ryder, Engineering Electronics, McGraw-Hill, Inc., NY, 1957.

'Cathode Ray' (M.G. Scroggie), "Negative Feedback and Non-Linearity,"
WW, Apr 1961, pp. 225-230; rewritten for transistors, WW, Oct 1978,
pp. 47-50.

Staff of the Department of Electrical Engineering: M.I.T., Magnetic
Circuits and Transformers, John Wiley and Sons, Inc., NY, 1958.

William H. Timbie, Basic Electricity for Communications, John Wiley
and Sons, Inc., NY, 1943.

[Kurt Strain]

Scott Frankland (audi...@ix.netcom.com) wrote:

: The solution is three-fold: (1) ELIMINATE the output transformer; (2)

: EXTEND the bandwidth into the MHz region (inasmuch as possible); and
: (3) apply negative LOOP FEEDBACK. This may be accomplished by means of
: multiple, paralleled output tubes. For example, given a feedback
: factor of 20db, sixteen parallel 6550's (converted to triode to reduce
: output impedance) will give a nominal damping factor of ten (as a
: rule, a damping factor of ten is considered to be just acceptable).

Sounds like the same argument used that solid state is the best way to
go. SS has all you say, and then some. Why bother with tubes at all?
The problem I have with technical arguments like this is that again it
tries to have it both ways: objective reasoning to explain a
subjective bias. Surely there is subjective reasoning for opting for
PP tubes themselves, one the transistor enthusiasts can easily reason
away as Scott did to SE amps here.

I am not going to say Scott can't make his 16 - 6550 OTL amp sound
glorious and measure well, too. But it strikes me odd that such a
strong technical reasoning in this authoritative post is presented
when stronger technical arguments exist for better measuring solid
state amps. You must admit to liking "euphony" as you say, or else
you are using that argument hypocritically. And the phrasing above,
"The solution is..." really ought to be modified to say "My solution
is..."

Personally I use both a standard 100W/ch 6550-based PP ultralinear amp
for a bass amp to 700 Hz and a 12W/ch parallel 2A3 SE amp for mids and
highs. Bass distortion is low and doesn't creep into the midrange,
and the power is adequate. The midrange is super transparent and
smooth, too. The damping factor of 4 is overkill for the mid/high
crossover network that effectively kills any amp's damping factor to a
limit of less than 2. The highs roll off above 30 KHz, zero global
feedback. There are no phase splitters, there is low distortion,
there is ultra soft clipping. It does "euphonically" cover the edgy
highs, just like some electrostatic speakers do, and few complain
about the great mids and pleasingly smooth highs they produce.

The speaker crossover is tailored to the response of the amp, a very
unusual approach. Why not? It's a system I'm building, not a
component. That's also why there's more than one amp in it, totally
different, but I find optimum for the whole. And it was made seemless
by tweaking on the speakers, amps, and even cables. It's not a mix
and match off-the- rack system, and that paradigm is something I think
is unfortunate in the high-end in general. Such amp-of-the-month club
members go crazy trying to dial in what was never optimized to fit
together, for all tastes.

The point is, it is too easy to throw out the strengths of one
approach just because something else has a few more strengths without
realizing that you can take the best from each approach and even
combine them. Such an attitude will assure unnecessary "obsolescence"
of good inventions. As it has happened before, and it becomes painful
to dig it back up.

If I want to add a subwoofer, be assured it will likely be driven by
solid state amps. Those amps have their place, too.

By the way, I was told that the most linear amplification device is a
short gate GaAsFET, which is not a square law device. Can we use it
in audio someday? Maybe not, it is way too noisy at audio
frequencies, with high 1/f noise. But, I can technically argue for
its use in certain parameters.

To conclude my comments, no one has yet done it all with one design
approach that objectively or subjectively makes everyone happy. To
state otherwise shows an agenda. So why is it there exists such a
struggle to try and convert others to their reasoning?

Kurt

[sdura...@aol.com]

In article <5804aq$5...@agate.berkeley.edu>, Scott Frankland
<audi...@ix.netcom.com> writes:

>Let's face it, gentlemen (and you lurking ladies, too), a truly
>transparent amplifier will not add ANY audible harmonics to the output
>(nor anything else not related to the input). As we used to say in the
>60's, feel free to roll your own--just remember, whatever you smoke is
>subject to color your blue skies gray. Of course, the opposite is
>sometimes true, as well. But in this case we are not talking about
>transparency so much as about euphony. And that is an entirely
>different subject altogether.

Whew, thanks Scott for your vigorous and complete appraisal of the
technical challenges facing SE triode amps.

What we have here is another situation in high end (and there are lots
of them), whereby preference flies in the face of demonstrable
technical shortcomings. And we've all been down the path of whose
preference is best.

As an SE triode listener, I have found (subjectively) a greater sense
of expression of subtle nuance in the reproduced music. This sense is
a very difficult thing to describe, let alone quantify. In sharing
listening experiences with other SE fans, there's some descriptive
commonality, such as greater sense of palpability, freedom from grain,
dimensionality, etc. By contrast, when you've listened to SE triodes
for a while, PP amps tend to sound hazy and ill-defined. But, when
all is said and done, the discussion seems to converge on this
intangible sense of greater musical expression to be had from SE
triode amps.

Clearly, not all listeners agree, else everyone would be using SE
triodes. Of those engineers who are familiar with the SE "sound," as
briefly described above, my question would be: is there a *dominant*
technical characteristic that creates this sound? Is it the lack of
damping? Is the the frequency response variation due to load
impedance? Compression? Transformer distortion?

Perhaps it's possible to build different PP amps and alter the design
elements to highlight each of these different characteristics and
determine which one is most responsible for the SE sound. Anyone
tried to do this?

Siegfried

"A lot of the problems in the world today stem from people not doing what
they should be doing..."

[Bill Thompson]

In article <582g07$6...@eyrie.graphics.cornell.edu>, sdura...@aol.com
wrote:

> In sharing listening experiences with other SE fans, there's some
> descriptive commonality, such as greater sense of palpability,
> freedom from grain, dimensionality, etc. By contrast, when you've
> listened to SE triodes for a while, PP amps tend to sound hazy and
> ill-defined.

> Of those engineers who are familiar with the SE "sound," as briefly

> described above, my question would be: is there a *dominant*
> technical characteristic that creates this sound?

I really believe that most of it is simply due to low-order even
harmonic distortion. The term "distortion" conjures up images of
fuzzy, grating, gritty, nasty sound but it doesn't necessarily have to
be so.

Low-feedback single ended amplifiers tend to generate even harmonic
distortion. Push-pull amplifiers generate symmetrical distortion
which contains only odd harmonics, not evens. I have noticed that
many single-ended fans also prefer low feedback, possibly because
lower feedback results in increased harmonic distortion.

In the commercial sound world, there is a device made by Aphex
Corporation called an "Aural Exciter" which takes an input signal,
cuts out frequencies below about 500Hz, and uses a diode as a
non-linear element to generate some even harmonics. The diode is not
driven very hard so the non-linearity is not extreme and the resulting
signal is not half-wave rectified, it is just slightly reshaped. If
you put a sine wave through this device, you see an output with a
slightly more "pointy" positive peak than the negative peak. This
smooth nonlinearity results in predominantly low-order even harmonics.
A small amount of this signal is added back to the input signal to
produce the "Aphex" sound.

I witnessed a demonstration of this system at the AES show several
years ago. A couple of dozen people were gathered in a room where a
master tape was played through a console and the Aphex effect could be
switched in and out. Virtually everyone there including me perceived
an added sense of detail, clarity, and sparkle with the effect
switched in. If you listened to the tape for awhile with the Aphex
effect and then switched it out, the result sounded suddenly "dull",
"hazy" and lacking in detail. You would swear that something vital
had been removed from the recording but obviously, in this case,
something pleasant was no longer being added. The perceived
difference and the reaction of the listeners to it seem strikingly
similar to your descriptions of "single-ended sound".

If you can ever get a demonstration of one of these Aphex boxes, you
might find it interesting.

"Aural Exciter" is a trademark of Aphex Corporation.

--
<- Bill Thompson - Ashly Audio Inc. Rochester, NY - b...@frontiernet.net ->
<- http://www.ashly.com bi...@ashly.com ->

[Scott Frankland]

Kurt Strain wrote:

> Sounds like the same argument used that solid state is the best way to
> go. SS has all you say, and then some.

Except for the tubes. <g>

> You must admit to liking "euphony" as you say, or else
> you are using that argument hypocritically.

I admit it, and freely.

> And the phrasing above,
> "The solution is..." really ought to be modified to say "My solution
> is..."

Well, true enough, Kurt. A large amp such as I describe might just
have problems that don't show up in an iddy bitty SETA job.

> The speaker crossover is tailored to the response of the amp, a very
> unusual approach. Why not? It's a system I'm building, not a
> component.

Sure, and I mentioned that in the post as "the systems approach". I
endorse it. That's how synergy is created. More power to you.

> The point is, it is too easy to throw out the strengths of one
> approach just because something else has a few more strengths without
> realizing that you can take the best from each approach and even
> combine them.

Yes, but can you sell the system AS A SYSTEM? The high-end has gotten
used to the idea of separates and the circuit specialists that stand
behind them. Perhaps that will change if the REVIEWERS begin to
endorse systems (the Avalon/Spectral/MIT combo comes to mind). It
seems to me that this ought to be part of the service that reviewers
provide to their paying subscribers.

> To conclude my comments, no one has yet done it all with one design
> approach that objectively or subjectively makes everyone happy. To
> state otherwise shows an agenda. So why is it there exists such a
> struggle to try and convert others to their reasoning?

I didn't intend for it to come off that way, Kurt, merely to raise the
issue of the much vaunted single-ended transparency to a fine point.
Besides, I only suggested the OTL SETA idea, I haven't actually done
it. My post should be looked at as "design work in progress". Of
course the final result will reflect my own preferences and biases.
How could it be otherwise unless I let the test bench decide when it's
right? For me, the test bench decides when it's wrong, but not right.

Regards,

~SF~

[Scott Frankland]

sdura...@aol.com wrote:
>
> As an SE triode listener, I have found (subjectively) a greater sense
> of expression of subtle nuance in the reproduced music. This sense is
> a very difficult thing to describe, let alone quantify.

But you see, it was not my intention to explain what SE does RIGHT,
but only to show the IMPEDIMENTS to neutrality and to true
transparency from a technical standpoint. What I did is nothing like a
complete statement of the issues, but merely a roster of the most
fundamental design problems.

> In sharing
> listening experiences with other SE fans, there's some descriptive
> commonality, such as greater sense of palpability, freedom from grain,
> dimensionality, etc. By contrast, when you've listened to SE triodes
> for a while, PP amps tend to sound hazy and ill-defined.

Careful, one shouldn't generalize these findings to the mode of
operation of the output stage. The problem could just as well lie
elsewhere--in the power supply, for example, or in the tube itself!

> Clearly, not all listeners agree, else everyone would be using SE

> triodes. Of those engineers who are familiar with the SE "sound," as

> briefly described above, my question would be: is there a *dominant*

> technical characteristic that creates this sound? Is it the lack of
> damping? Is the the frequency response variation due to load
> impedance? Compression? Transformer distortion?

You forgot to mention negative-cycle curvature, and the even-order
distortion products that result from it. Most of the SE proponents
(and I am not necessarily against SE operation) point to the complete
absence of crossover distortion in SE amps, and to the deletion of the
phase-splitter stage. Since no one has brought this up as yet, I won't
bother to argue for or against.

> Perhaps it's possible to build different PP amps and alter the design
> elements to highlight each of these different characteristics and
> determine which one is most responsible for the SE sound. Anyone
> tried to do this?

Well, that's pretty much what I suggested in my conclusion. Build an
OTL SETA. That will tell you immediately what the contribution of the
transformer might be. Next, vary the feedback (taking pains to also
optimize the compensation). This will not only increase the even-order
distortion, but will reduce the damping factor, as well.

Understand that I have made no value judgments. No one is saying that
perfect accuracy can be obtained, and no one is saying that euphonic
products are not desirable to some degree in light of this fact. I
have said nothing as yet about euphonics, other than that they do, in
many ways, contribute to what people define as the "SE sound".

~SF~

[Scott Frankland]

Bill Thompson wrote:
>
> Low-feedback single ended amplifiers tend to generate even harmonic
> distortion. Push-pull amplifiers generate symmetrical distortion
> which contains only odd harmonics, not evens.

Push-pull amps also produce even-order harmonics, but they are
necessarily less dominant. It would require a perfectly balanced
push-pull amp with ideal devices to produce pure odd-order distortion.

> I have noticed that
> many single-ended fans also prefer low feedback, possibly because
> lower feedback results in increased harmonic distortion.

Lower feedback NECESSARILY results in increased harmonic distortion
provided that increasing the feedback does not overload the first
stage or produce instability.

> In the commercial sound world, there is a device made by Aphex
> Corporation called an "Aural Exciter" which takes an input signal,
> cuts out frequencies below about 500Hz, and uses a diode as a
> non-linear element to generate some even harmonics. The diode is not
> driven very hard so the non-linearity is not extreme and the resulting
> signal is not half-wave rectified, it is just slightly reshaped. If
> you put a sine wave through this device, you see an output with a
> slightly more "pointy" positive peak than the negative peak. This
> smooth nonlinearity results in predominantly low-order even harmonics.
> A small amount of this signal is added back to the input signal to
> produce the "Aphex" sound.

Reshaping the waveform on one end only can produce a dizzying array of
special effects, most of which will be euphonic unless sharp kinks are
introduced. One of the most compelling reasons for owning tube
equipment is that the end-user is allowed to produce similar effects
for himself, as each brand of tube has its own characteristic sound
that will either enhance or detract from the perceived reality
according to the taste of the listener.

~SF~

[Jeff Bernhard]

Bill Thompson (b...@frontiernet.net) wrote:
: I really believe that most of it is simply due to low-order even

: harmonic distortion. The term "distortion" conjures up images of
: fuzzy, grating, gritty, nasty sound but it doesn't necessarily have to
: be so.

[Deletia]
: In the commercial sound world, there is a device made by Aphex
: Corporation called an "Aural Exciter" [...]
[Deletia]
: I witnessed a demonstration of this system at the AES show several

: years ago. A couple of dozen people were gathered in a room where a
: master tape was played through a console and the Aphex effect could be
: switched in and out. Virtually everyone there including me perceived
: an added sense of detail, clarity, and sparkle with the effect
: switched in. If you listened to the tape for awhile with the Aphex
: effect and then switched it out, the result sounded suddenly "dull",
: "hazy" and lacking in detail. You would swear that something vital
: had been removed from the recording but obviously, in this case,
: something pleasant was no longer being added. The perceived
: difference and the reaction of the listeners to it seem strikingly
: similar to your descriptions of "single-ended sound".

I found this the most enlightening post on the subject of euphonic
distortion that I've ever read. Perhaps this is because it described
the effects of the distortion on perception rather than saying
something like, "Even order products are euphonic" or "SE afficionados
prefer their amps for the euphonic distortions generated". These
statements may be true but don't give me a 'feel' of the effects. This
is similar to my experience (sometimes) of live music: At first the
sound may see less 'exciting' than recorded music -- dull even. Then
the perception of 'dull' changes (as I acclimate) to 'pristine', and
the natural sound of woodwinds and strings and other instruments
becomes utterly seductive. My question: Do those who love euphonic
products prefer them over live music on a sonic basis? --

Jeffrey Bernhard Concurrent Computer Corp.
Jeff.B...@mail.hcsc.com Voice: (954) 973-5496 Fax: (954) 977-5580
*** The opinions expressed herein are mine, not those of my employer! ***

[sdura...@aol.com]

[Moderators Note - Somewhat against my better judgement I have
allowed one of Siegfried's "little questions" to be posted.
I hope the audio gods don't judge me too harshly -- bt]

In article <584ql8$i...@canyon.sr.hp.com>, Scott Frankland
<audi...@ix.netcom.com> writes:

>You forgot to mention negative-cycle curvature, and the even-order
>distortion products that result from it. Most of the SE proponents
>(and I am not necessarily against SE operation) point to the complete
>absence of crossover distortion in SE amps, and to the deletion of the
>phase-splitter stage. Since no one has brought this up as yet, I won't
>bother to argue for or against.

Actually, I brought up the deletion of the phase-splitter some time
ago in r.a.h-e only to face general ridicule. Could you explain how
the lack of a phase-splitter could result in sound that is more
transparent than wide bandwidth PP amps that would not (presumably)
suffer from phase distortion?

Siegfried

[Sig .sig deleted with extreme prejudice - get a new one -- bt]

[Scott Frankland]

sdura...@aol.com wrote:

> Could you explain how
> the lack of a phase-splitter could result in sound that is more
> transparent than wide bandwidth PP amps that would not (presumably)
> suffer from phase distortion?

In my original post to this thread I explained how bass distortion, so
prominent in SETA's that use no feedback, will add energy to midrange
harmonics. Since the distortion produces are low-order in triodes (and
therefore benign), this can increase the subjective sense of
transparency in triode SETA's (unless the added energy becomes
excessive). This may very well be the essence of the explanation you
seek.

Let's go further, however. It's important to realize that the
phase-splitter can be designed to provide GAIN, in addition to
providing a pair of balanced signals with which to drive the output
stage. If the PP amp is designed without feedback, then the SAME
number of gain stages can be used as in the SETA! The only difference
is that in the PP amp the signals will be balanced, and there will be
virtually no even-order distortion products.

This, I think, is the key. The odd-order residue that then remains in
the PP amp is more foreign to the ear than is a smoothly converging
pattern of even + odd harmonics (as in the SETA). This conclusion is
supported by the fact that, when the ear overloads, it produces the
same pattern of smoothly converging even + odd harmonics as in the
SETA.

That is why I believe it is essential to reduce the odd order residue
in PP amps to below the threshold of audibility. That's the key for
push-pull. If that can't be accomplished, then I think that many
people may find that the SETA sounds more pleasing. Thus I conclude
that negative feedback, properly applied, is a necessity in PP
amplifiers.

By the same token, however, reducing even-order distortion products in
the SETA should likewise increase the TRUE transparency or resolving
power. There is thus room to improve in both technologies. That is the
real point of my post. It is difficult to apply feedback effectively
around SETA's, however, due to the restricted open-loop bandwidth.
Thus the desirability for an OTL SETA.

These comments are restricted to tube amplifiers in the belief that
tubes add palpability to the sound of music that is not related to
distortion products. I am willing to concede that this palpability is
quite likely due to electrode resonances. So what, if they increase
the sense of realism for many listeners?

Is this not the secret to the mystery of why Audio Research offers
similar products in your choice of either tubes or FET's?

~SF~

[Scott Frankland]

Jeff Bernhard wrote:
[snip - moderator bt]
>...Perhaps this is because it described

> the effects of the distortion on perception rather than saying

> something like, "Even order products are euphonic" ....

You have the makings of a fine subjectivist me boy! Harry Pearson,
paging Harry Pearson!

[snip -- moderator bt]

> My question: Do those who love euphonic
> products prefer them over live music on a sonic basis? --

This is a philosophical question. The question reduces to one of:
"What is human sensibility, and how is it best served?" Care to follow
up on it?

~SF~

[Richard D Pierce]

In article <587lcs$h...@canyon.sr.hp.com>, <sdura...@aol.com> wrote:
>[Moderators Note - Somewhat against my better judgement I have
>allowed one of Siegfried's "little questions" to be posted.
>I hope the audio gods don't judge me too harshly -- bt]

>>You forgot to mention negative-cycle curvature, and the even-order

>>distortion products that result from it. Most of the SE proponents
>>(and I am not necessarily against SE operation) point to the complete
>>absence of crossover distortion in SE amps, and to the deletion of the
>>phase-splitter stage. Since no one has brought this up as yet, I won't
>>bother to argue for or against.

>Actually, I brought up the deletion of the phase-splitter some time
>ago in r.a.h-e only to face general ridicule.

No, you erroneously insisted that phase splitters caused problems with
"switching" between positive and negative cycles, something which
simply does not occur at all in phase splitters. Despite the technical
correction, you persisted. It was that incorrect technical assertion
that was held to public ridicule, not you personally.

Justy thought I would clarify that point. Again.

>Could you explain how
>the lack of a phase-splitter could result in sound that is more
>transparent than wide bandwidth PP amps that would not (presumably)
>suffer from phase distortion?

If you are going to appropriate the use of the term phase disttortion,
then you will have to face the consequennces of using a well defined
term in widespread use.

As such, the phase distortion you alude to DOES NOT EXIST in
phase-splitter based devices. In fact, phase distortion is a trivially
measurable quantity, and what exists of it is not produced by the
phase splitter or lack thereof, but rather is a complimentary property
of the bandwidth and frequency response of the system. One can
trivially show that if an amplifier that did not use a phase splitter
had a narrower bandwidth and worse frequency response, its phase
response would be demonstrably worse than that of an amplifier that
utilized a phase splitter that had a wider and fl;atter frequency
response.

The problem here, I suspect, is that you are focusing on the NAME of
the device, NOT the PROPERTY of the device. In that sense, a phase
splitter is a misnomer. You have seemed in the past to assert (or
assume) that the action of the device was to take the positive half of
the waveform and send it off to one side of the amplifer, and then
take the other half of the waveform and send it off to the other half.
If this is the assumption, that assumption is dead wrong.

In fact, what the device does is takes the ENTIRE waveform and sends
it to one side of the amp, and SIMULTANEOUSLY takes the SAME entire
waveform, inverted 180 degrees and sends it off to the other side.
There is no "phase splitting" in the sense of the word you are using
it. In all the designs I am familiar with, the phase splitter is, in
fact, a single-ended class A amplifier with balanced outputs. Now, the
simple examples may have some problems in that there is effectively
different source impedances for the positive and negative sides, but
a) that's a knowable problems and can be overcome and b) if anything,
makes the system behave more like an SE amplifier because of the
non-symmtery it introduces, analogous to the non-symmetry in SE
amplifiers.

Get off the "phase splitter" hobby horse. There's more than enough
reasons to explain the differences without inventing something
entirely new.

--
| Dick Pierce |
| Loudspeaker and Software Consulting |
| 17 Sartelle Street Pepperell, MA 01463 |
| (508) 433-9183 (Voice and FAX) |

[Steve Jones, contractor]

Jeff Bernhard wrote:
>
> My question: Do those who love euphonic
> products prefer them over live music on a sonic basis? --

In a word no.

This comes back to a rhetorical question I asked on rahe a couple of
weeks ago. The question was:

Stereo may require a "dithering agent".

Some netizens jumped right in to correct me. They were right in thier
explainations, BUT, in MY question, the key operative word was
"Stereo".

Let me explain. Stereo is in the preception. Where in nature do two
different sound sources produce a "phantom image"? Can the phantom
image be accurately measured? To what tolerance? Inter-aural
crosstalk (comb filtering) pretty much does away with flat frequency
response (for the phantom image) at the listeners ears in the 2k and
above range. Also stereo mid-bass PERCEPTION is not at all accurate
if live music is your reference. In this regard Surround Sound can be
much more presuasive.

IF the distortion artifacts are of a decreasing (as harmonic order
rises) nature, the perceived 'stereo' sound may be a closer simulation
of the 'live' experience. Henry Pasternack relayed an interesting
E-Mail from Jean Hiraga about the spectral content of this applied
'consonant' distortion a few months back. Suggested levels for the (if
I remember correctly) of the 2nd through 5th in relation to the main
signal level. Mr. Hiraga started this SE conundrum in the western
world when he brought the Japanese Triode fad to France circa late
1970's. Is this the end of discussion. No. Hopefully the starting of a
longer one. No offense to senisiblities intended.

My US$0.03
-Steve Jones

BTW, I love live music. Get every chance I can. And lately have been
lamenting on the audio chatlines the observation that: 'Stereophile'
by the numbers home reproduction is at best 'missing something'
compared to the live experience.

Reminds me of going to an outside concert (Nanci Griffith a couple of
years ago) with a "reviewer" from Positive Feedback Gagazine. His
audiophile system was much ballyhooed in his writings and indeed
sounded very good. But live, NOT! Anyway, after the concert, he said
he would never go to another concert like that again. His stereo
sounded better. I've seen Nanci three times since.

Do people who recognize the fact that, yes SET's ADD distortion to the
signal, prefer this distortion over 'the live sound" Not for me. It
simply is FAR more realistically protrayed (when done right), than
what the store bought stuff will do.

[Donald J. North]

Hi

I think it should be noted quite clearly that single-ended amplifiers
are not inherently plagued by a high output impedance or that they are
the only class of operation with this issue. Push-pull tube amps also
suffer from this problem if paralleled output devices are not used,
for the same transformer impedance, or if negative loop feedback is
not used. A good example of this is the measured performance of the
VAC amplifiers when they were tested in Stereophile. Both the triode
and beam tetrode models had an output impedance of over 2 ohms!
Another example is the new MESA Baron tube amplifier, where Audio
magazine measured an output impedance of over 8 ohms!

If loop feedback were used, then this output impedance issue could be
lowered, in light of SF's mention of the difficulties with feedback in
SE triodes.

This is not to say that a loudspeaker cannot be designed to perform
optimally when connected to a SE triode amplifier. If the speaker were
designed for specific use with high output impedance amplifiers, both
PP and SE, then the damping and frequency response shaping issues
could be solved. Then a composite "system" frequency response could be
as those for speakers designed for use with low output impedance amps
and used with these amps.

Donald North

[Jeff Bernhard]

Richard D Pierce (DPi...@world.std.com) wrote:
: [...] You have seemed in the past to assert (or

: assume) that the action of the device was to take the positive half of
: the waveform and send it off to one side of the amplifer, and then
: take the other half of the waveform and send it off to the other half.
: If this is the assumption, that assumption is dead wrong.

: In fact, what the device does is takes the ENTIRE waveform and sends
: it to one side of the amp, and SIMULTANEOUSLY takes the SAME entire
: waveform, inverted 180 degrees and sends it off to the other side.
: There is no "phase splitting" in the sense of the word you are using
: it. In all the designs I am familiar with, the phase splitter is, in

: fact, a single-ended class A amplifier with balanced outputs. [...]

: Get off the "phase splitter" hobby horse. There's more than enough

: reasons to explain the differences without inventing something
: entirely new.

I found this explanation to not only clarify this issue, but reminds
me that often in the high-end press, a reviewer will hear something
and make an 'educated' guess as to the cause, e.g,. "a phase anomaly
somewhere around 80 Hz." or somesuch. The manufacturer never replies
to this claim, and readers may assume the reviewer is right, and some
sort of genius for diagnosing by ear what an engineer could only
determine by measurement. I think many of the misunderstandings and
much of the snake oil is because such pseudo-technical claims are
allowed to go unchallenged, and I appreciate that r.a.h-e is a forum
in which more sane discourse can (sometimes) prevail. (It would be
nice if reviewers didn't offer their 'explanations', which IMHO have
little chance of being correct and generally detract from the rest of
the review.) --

[Steve Jones, contractor]

Scott Frankland wrote:

> In my original post to this thread I explained how bass distortion, so
> prominent in SETA's that use no feedback, will add energy to midrange
> harmonics. Since the distortion produces are low-order in triodes (and
> therefore benign), this can increase the subjective sense of
> transparency in triode SETA's (unless the added energy becomes
> excessive).

Snip

> This, I think, is the key. The odd-order residue that then remains in
> the PP amp is more foreign to the ear than is a smoothly converging
> pattern of even + odd harmonics (as in the SETA). This conclusion is
> supported by the fact that, when the ear overloads, it produces the
> same pattern of smoothly converging even + odd harmonics as in the
> SETA.
>
> That is why I believe it is essential to reduce the odd order residue
> in PP amps to below the threshold of audibility. That's the key for
> push-pull. If that can't be accomplished, then I think that many
> people may find that the SETA sounds more pleasing. Thus I conclude
> that negative feedback, properly applied, is a necessity in PP
> amplifiers.

I've had seeming good results in un-balancing the output tubes in my
current P-P amp (with no NFB excepting cathode degeneration) by about
5mA. My rational was to:

#1) Separate the AC crossing from the signal zero crossing
(an inexact anology: similar to the Burr Brown Co-Linear DACs that
combine multi and single bit archecture to move noise away from
the LSB.)
#2) ADD primarily 2nd and higher even order distortion to help in masking
the naked odd order stuff.

Result? A sound closer to real (for me).

> By the same token, however, reducing even-order distortion products in
> the SETA should likewise increase the TRUE transparency or resolving
> power.

Usually done by running the triode at near its maximum plate voltage,
near maximum plate dissipation, with a higher than "quoted" plate
load. Pentodes (including beam power types) seem to like a lower B+
with alot of current (near max plate disipation) and a lower than
"quoted" plate load. This is to minimize the 3rd and higher odd
harmonics.

-My next amp is a SET (in the works). BTW, Scott, any (re)sources for
information on high frequency filament supplies for a DHT (5VAC @
1.2A, ~170khz -solid state will do)?

> These comments are restricted to tube amplifiers in the belief that
> tubes add palpability to the sound of music that is not related to
> distortion products. I am willing to concede that this palpability is
> quite likely due to electrode resonances. So what, if they increase
> the sense of realism for many listeners?

In 100% agreement that Tubes ADD palpability to stereo that even the
best solid state misses. Personally I WOULD relate it to distortion
products (or more precisely 'distortion spectra'). But, why quibble (
how many angels CAN fit on the head of a pin)? It's all there in the
listening!

Great discussion.

Enjoy!
-Steve Jones

[Stewart Pinkerton]

no...@alumnae.caltech.edu (Donald J. North) writes:

>This is not to say that a loudspeaker cannot be designed to perform
>optimally when connected to a SE triode amplifier. If the speaker were
>designed for specific use with high output impedance amplifiers, both
>PP and SE, then the damping and frequency response shaping issues
>could be solved. Then a composite "system" frequency response could be
>as those for speakers designed for use with low output impedance amps
>and used with these amps.

So what exactly was your point? The fact that something CAN be done
does not mean it HAS been done. As far as I know, no such 'system' has
EVER been produced, hence one is left with the situation that high
output impedance in amplifiers is a BAD thing and inexcusable, since
it's perfectly possible to produce highly linear amplifiers which do
not suffer this problem and are therefore able to drive ANY speaker
without affecting its frequency response (or distorting the input
signal).

--

Stewart Pinkerton | If you can't measure what you're making,
A S P Consulting | how do you know when you've got it made?
(44) 1509 880112 |

"I canna change the laws o' physics" - the other Scotty

[Stewart Pinkerton]

"Steve Jones, contractor" <sjo...@vcd.hp.com> writes:

>Reminds me of going to an outside concert (Nanci Griffith a couple of
>years ago) with a "reviewer" from Positive Feedback Gagazine. His
>audiophile system was much ballyhooed in his writings and indeed
>sounded very good. But live, NOT! Anyway, after the concert, he said
>he would never go to another concert like that again. His stereo
>sounded better. I've seen Nanci three times since.

I wish you'd named names here. We could have a new dictionary
reference for asshole...............................

[ Ah, our dear Stewart. Never one to mince words. -- jwd ]

>Do people who recognize the fact that, yes SET's ADD distortion to the
>signal, prefer this distortion over 'the live sound" Not for me. It
>simply is FAR more realistically protrayed (when done right), than
>what the store bought stuff will do.

Yep. I keep trying and buying and tweaking and freaking, but I have
not yet heard a sound system which comes even into the same county as
a symphony orchestra in full flight. Chamber music? Well, maybe with a
following wind, but NOT a full orchestra!

[sdura...@aol.com]

In article <588eqq$t...@agate.berkeley.edu>, Scott Frankland
<audi...@ix.netcom.com> writes:

> In my original post to this thread I explained how bass distortion,
> so prominent in SETA's that use no feedback, will add energy to
> midrange harmonics. Since the distortion produces are low-order in
> triodes (and therefore benign), this can increase the subjective
> sense of transparency in triode SETA's (unless the added energy

> becomes excessive). This may very well be the essence of the
> explanation you seek.

It seems to me, then, that there should be an ideal characteristic and
amount of bass distortion causing these midrange harmonics that
increase the subjective sense of transparency. Two questions:

Are there any indications in the scientific literature that support
the notion of added harmonics creating a sense of "transparency" (NOT
euphonics or "pleasing", but transparent in the sense that the subject
has a heightened aural sense of the original soundfield)?

If so, why don't engineers design around this protocol rather than the
idealized sense of linear accuracy that pervades the industry?

> That is why I believe it is essential to reduce the odd order
> residue in PP amps to below the threshold of audibility. That's the
> key for push-pull. If that can't be accomplished, then I think that
> many people may find that the SETA sounds more pleasing. Thus I
> conclude that negative feedback, properly applied, is a necessity in
> PP amplifiers.

Yes, I see your point. Would you also say that feedback doesn't do
much to REDUCE odd-order harmonics in SETAs?

> By the same token, however, reducing even-order distortion products
> in the SETA should likewise increase the TRUE transparency or
> resolving power.

As I refereed to above, perhaps there's a point below which distortion
should not be pushed?

> These comments are restricted to tube amplifiers in the belief that
> tubes add palpability to the sound of music that is not related to
> distortion products. I am willing to concede that this palpability
> is quite likely due to electrode resonances. So what, if they
> increase the sense of realism for many listeners?

But aren't electrode resonances just another form of distortion? How
do they differ from regular distortion? Or, put another way, what IS
responsible for distortion in amplifiers, regardless of topology or
anything else? Some of it must be thermal noise, some of it
resonance, etc. Of, say, 0.1% THD, what operational elements of the
amplifier make up that distortion and how is the distortion pie split
up between the various causes?

--

[Scott Frankland]

Steve Jones, contractor wrote:
>
> I've had seeming good results in un-balancing the output tubes in my
> current P-P amp (with no NFB excepting cathode degeneration) by about
> 5mA. My rational was to:

[snip]

> Result? A sound closer to real (for me).

I have no problem with this approach, either, if the end justifies the
means. In fact, there may be a better way yet to unbalance the
amplifier--use asymmetric feedback. This bends the transfer curve in
just the right way so as to produce a distortion spectrum not unlike
that of a SETA.

> -My next amp is a SET (in the works). BTW, Scott, any (re)sources for
> information on high frequency filament supplies for a DHT (5VAC @
> 1.2A, ~170khz -solid state will do)?

I'd be willing to bet that Pete Goudrea can answer this
question. Pete?

[ cutting -- rgd ]

> In 100% agreement that Tubes ADD palpability to stereo that even the
> best solid state misses. Personally I WOULD relate it to distortion
> products (or more precisely 'distortion spectra').

Well, I'm planning to test this theory by designing a very low
distortion (i.e., low wideband distortion) tube amp. I intend to do it
for both a pentode PP and an OTL SETA (triode).

> Great discussion.

Thanks!

~SF~

[Scott Frankland]

sdura...@aol.com wrote:
> In article <588eqq$t...@agate.berkeley.edu>, Scott Frankland
> <audi...@ix.netcom.com> writes:

[ cut -- rgd ]

> It seems to me, then, that there should be an ideal characteristic =

> and amount of bass distortion causing these midrange harmonics that

> increase the subjective sense of transparency. =

That is, if you are willing to tolerate bass distortion, which I am =
not.

> Two questions: =

> Are there any indications in the scientific literature that support
> the notion of added harmonics creating a sense of "transparency" (NOT
> euphonics or "pleasing", but transparent in the sense that the subject
> has a heightened aural sense of the original soundfield)?

In previous posts I have referenced Olson, Winckel, Ladner, and
Shorter. =

I suggest you start there. The sense of transparency AS transparency
is to some extent in the ear of the beholder. What will be most
affected is the timbre of instruments. The distortion spectrum as
produced by many SETA's apparently makes instruments more palpable, =
immediate, or "there" for many listeners. Other listeners find that =
the presentation is due to artifacts. =

There is a passage from the legendary *Radiotron Designer=92s
Handbook*, = under the subhead: =93The ear as a judge of fidelity=94,
that is appropriat= e =

here: =93It is common practice to regard the ear as the final judge of
= fidelity, but this can only give a true judgment when the listener
has = acute hearing, a keen ear for distortion, and is not in the
habit of = listening to distorted music. A listener with a keen ear
for distortion = can only cultivate this faculty by making frequent
direct comparisons = with the original music in the concert hall [4th
ed., 1953, pg. 632.].=94 = (Shades of HP!) =

That, apart from the enjoyment, is one of the best reasons to attend =
live concerts on a regular basis. Even if one chooses to ignore = the
paradigm known as the absolute sound=97-i.e., that our reproduced =
music should imitate as closely as possible what we hear live-=97we =
can still benefit by making the comparison. That is because, in so =
doing, we will learn to distinguish distortions in reproduced music.

SETA's are no panacea. Even the best SETA can be shown to have
problems, and virtually any listener can, with persistence, be trained
to hear the artifacts AS artifacts rather than as =

"transparency". It just so happens that for many listeners the willing
suspension of disbelief occurs more quickly and easily = with a good
SETA in the system. I can't argue with that.

> If so, why don't engineers design around this protocol rather than =

> the idealized sense of linear accuracy that pervades the industry?

They do--in SETA's. The SETA movement is to a large extent a movement
= away from accuracy as a criterion and toward beauty, emotion, and
sheer = involvement. It creates the strongest line of demarcation yet
between = tubes and transistors. =

> > That is why I believe it is essential to reduce the odd order
> > residue in PP amps to below the threshold of audibility. That's the
> > key for push-pull. If that can't be accomplished, then I think that
> > many people may find that the SETA sounds more pleasing. Thus I
> > conclude that negative feedback, properly applied, is a necessity in
> > PP amplifiers.

> Yes, I see your point. Would you also say that feedback doesn't do
> much to REDUCE odd-order harmonics in SETAs?

You will recall from the recent thread Re: Distortion and Feedback
that feedback does not discriminate between the orders. You are
confusing feedback with selective cancellation, such as occurs in PP
amps. =

Feedback will reduce both odd as well as even-order distortion in any
amplifier in which feedback is properly applied. The problem is that =
it is difficult to obtain feedback effectively in SETA's due to their
= limited open-loop bandwidth. That is why I suggested an OTL SETA, =
which would not suffer so much from this problem.

> > By the same token, however, reducing even-order distortion products
> > in the SETA should likewise increase the TRUE transparency or
> > resolving power.

> As I refereed to above, perhaps there's a point below which distortion
> should not be pushed?

Well, I think it should be eliminated, but that doesn't seem to do it
for me either, for reasons which I can't explain just yet. I think
that, if there is in fact distortion present, it should be smoothly
converging, and should contain even as well as odd products (for the
reasons stated in a previous post). =

> > These comments are restricted to tube amplifiers in the belief that
> > tubes add palpability to the sound of music that is not related to
> > distortion products. I am willing to concede that this palpability
> > is quite likely due to electrode resonances. So what, if they
> > increase the sense of realism for many listeners?

> But aren't electrode resonances just another form of distortion? How
> do they differ from regular distortion? Or, put another way, what IS
> responsible for distortion in amplifiers, regardless of topology or

> anything else? =

Anything present at the output (other than gain) which is not present
at the input, is distortion. Electrode resonances are signal-induced,
= caused by electrons impacting the anode plate and bouncing off and =
around inside the tube like BB's (thus demonstrating the particle =
nature of electrons!). Tube resonance is thus an electromechanical =
effect (and don't forget the airborne acoustically-induced resonance =
as well). The tube thus gets it from within and from without. =

Harmonic distortion is a purely electrical effect due to curvature =
of the device conductance characteristics. IM distortion is due to =
sum and difference products between the spurious harmonic products. =
Noise can modulate these products, further altering the perceived =
result. Resonance adds spice to the mix. :-0 =

> Some of it must be thermal noise, some of it
> resonance, etc. Of, say, 0.1% THD, what operational elements of the
> amplifier make up that distortion and how is the distortion pie split
> up between the various causes?

It will be different for every amplifier. That much I can state
unequivocally. I recommend that you make a study of the Stereophile =
back issues relating to amplifiers. Look at the distortion plots for =
distortion vs. load vs. power output vs. frequency, etc. You will =
find that it is not a simple question to answer. =

~SF~

[Scott Frankland]

sdura...@aol.com wrote:

> In article <588eqq$t...@agate.berkeley.edu>, Scott Frankland
> <audi...@ix.netcom.com> writes:

[ cut -- rgd ]

> It seems to me, then, that there should be an ideal characteristic

> and amount of bass distortion causing these midrange harmonics that
> increase the subjective sense of transparency.

That is, if you are willing to tolerate bass distortion, which I am not.

> Two questions:

> Are there any indications in the scientific literature that support
> the notion of added harmonics creating a sense of "transparency"
> (NOT euphonics or "pleasing", but transparent in the sense that the
> subject has a heightened aural sense of the original soundfield)?

In previous posts I have referenced Olson, Winckel, Ladner, and

Shorter. I suggest you start there. The sense of transparency AS

transparency is to some extent in the ear of the beholder. What will
be most affected is the timbre of instruments. The distortion spectrum
as produced by many SETA's apparently makes instruments more palpable,

immediate, or "there" for many listeners. Other listeners find that

the presentation is due to artifacts.

There is a passage from the legendary *Radiotron Designer's Handbook*,
under the subhead: "The ear as a judge of fidelity", that is
appropriate here: "It is common practice to regard the ear as the
final judge of fidelity, but this can only give a true judgment when
the listener has acute hearing, a keen ear for distortion, and is not
in the habit of listening to distorted music. A listener with a keen
ear for distortion can only cultivate this faculty by making frequent
direct comparisons with the original music in the concert hall [4th
ed., 1953, pg. 632.]. (Shades of HP!)

That, apart from the enjoyment, is one of the best reasons to attend

live concerts on a regular basis. Even if one chooses to ignore the
paradigm known as the absolute sound (i.e., that our reproduced music
should imitate as closely as possible what we hear live) we can still
benefit by making the comparison. That is because, in so doing, we

will learn to distinguish distortions in reproduced music.

SETA's are no panacea. Even the best SETA can be shown to have
problems, and virtually any listener can, with persistence, be trained

to hear the artifacts AS artifacts rather than as "transparency". It

just so happens that for many listeners the willing suspension of

disbelief occurs more quickly and easily with a good SETA in the

system. I can't argue with that.

> If so, why don't engineers design around this protocol rather than

> the idealized sense of linear accuracy that pervades the industry?

They do--in SETA's. The SETA movement is to a large extent a movement

away from accuracy as a criterion and toward beauty, emotion, and

sheer involvement. It creates the strongest line of demarcation yet
between tubes and transistors.

>> That is why I believe it is essential to reduce the odd order
>> residue in PP amps to below the threshold of audibility. That's
>> the key for push-pull. If that can't be accomplished, then I think
>> that many people may find that the SETA sounds more pleasing. Thus
>> I conclude that negative feedback, properly applied, is a necessity
>> in PP amplifiers.

> Yes, I see your point. Would you also say that feedback doesn't do
> much to REDUCE odd-order harmonics in SETAs?

You will recall from the recent thread Re: Distortion and Feedback
that feedback does not discriminate between the orders. You are
confusing feedback with selective cancellation, such as occurs in PP
amps.

Feedback will reduce both odd as well as even-order distortion in any

amplifier in which feedback is properly applied. The problem is that

it is difficult to obtain feedback effectively in SETA's due to their

limited open-loop bandwidth. That is why I suggested an OTL SETA,

which would not suffer so much from this problem.

>> By the same token, however, reducing even-order distortion products
>> in the SETA should likewise increase the TRUE transparency or
>> resolving power.

> As I refereed to above, perhaps there's a point below which
> distortion should not be pushed?

Well, I think it should be eliminated, but that doesn't seem to do it
for me either, for reasons which I can't explain just yet. I think
that, if there is in fact distortion present, it should be smoothly
converging, and should contain even as well as odd products (for the
reasons stated in a previous post).

>> These comments are restricted to tube amplifiers in the belief that

>> tubes add palpability to the sound of music that is not related to
>> distortion products. I am willing to concede that this palpability
>> is quite likely due to electrode resonances. So what, if they
>> increase the sense of realism for many listeners?

> But aren't electrode resonances just another form of distortion?
> How do they differ from regular distortion? Or, put another way,
> what IS responsible for distortion in amplifiers, regardless of
> topology or anything else?

Anything present at the output (other than gain) which is not present

at the input, is distortion. Electrode resonances are signal-induced,

caused by electrons impacting the anode plate and bouncing off and

around inside the tube like BB's (thus demonstrating the particle

nature of electrons!). Tube resonance is thus an electromechanical

effect (and don't forget the airborne acoustically-induced resonance

as well). The tube thus gets it from within and from without.

Harmonic distortion is a purely electrical effect due to curvature of
the device conductance characteristics. IM distortion is due to sum
and difference products between the spurious harmonic products. Noise

can modulate these products, further altering the perceived

result. Resonance adds spice to the mix. :-0

> Some of it must be thermal noise, some of it resonance, etc. Of,

> say, 0.1% THD, what operational elements of the amplifier make up
> that distortion and how is the distortion pie split up between the
> various causes?

It will be different for every amplifier. That much I can state
unequivocally. I recommend that you make a study of the Stereophile

back issues relating to amplifiers. Look at the distortion plots for

distortion vs. load vs. power output vs. frequency, etc. You will find

that it is not a simple question to answer.

~SF~

[sdura...@aol.com]

In article <58789u$d...@canyon.sr.hp.com>, je...@amber.hcsc.com (Jeff
Bernhard) writes:

> My question: Do those who love euphonic
>products prefer them over live music on a sonic basis? --

While I prefer non-linear components in certain cases, I don't perceive
them as being "euphonic." Let me explain why.

The idea of euphonic or pleasing to the ear requires a value judgment on
the aesthetic nature of what one is hearing. Proper audio reproduction
should require no aesthetic judgement - it should merely be an assessment
of how closely the reproduction matches the person's perception of what
live music sounds like to them.

I've yet to hear ANY audio reproduction sound like live music. Hence, by
definition, no audio product is preferable to live music. But when trying
to achieve a sound that is as close to live as possible, I've found that
certain non-linear products get me closer to that sense of live sound.

I understand that our ears are not the same and that there may be some out
there who find that the MOST linear products best replicate live sound.
But if you take into consideration how we hear differently, I can't see
how SO many people naturally fall into the category of preferring linear
products. Many of these people really haven't taken the time to fully
explore their hearing response and matching it to the best audio products
for their hearing. They merely accept linear is correct because it is
mathematically proper and let it go at that.

Are you one of these people? Have you taken the time to test yourself and
truly MAKE SURE that linear products offer the closest sense to your
perception of live music?

[Scott Frankland]

sdura...@aol.com wrote:
>
> The idea of euphonic or pleasing to the ear requires a value judgment on
> the aesthetic nature of what one is hearing. Proper audio reproduction
> should require no aesthetic judgement - it should merely be an assessment
> of how closely the reproduction matches the person's perception of what
> live music sounds like to them.

Your definition of "proper audio reproduction" is interesting, but it
fails to distinguish between audio criticism and the pure listening
experience, which I feel is an important distinction. When we listen
to music played back, we do not make "assessments" unless we are in
the critical mode. The only way to become involved with the music is
to drop the critical mode, i.e., to willingly suspend our disbelief.
In so doing, we enter the realm of pure aesthetic experience. These
two modes, therefore, are to a large extent self-defeating.

This is not to say that each mode does not each have its place. The
idea is to build one's system in such a way as to make it easy to
willingly suspend one's disbelief. That process requires critical
judgment. The important thing, in any case, is to enter the realm of
aesthetic experience with as little hindrance as possible.

Whether the music sounds live or not is a separate issue. Live music
is the standard because it sounds *better* than reproduced music. This
is not to say that it might not be possible to someday reverse these
roles. The problem here is akin to building a better sounding violin
than a Stradivarius.

~SF~

[Steve Jones, contractor]

Scott Frankland wrote:

> In previous posts I have referenced Olson, Winckel, Ladner, and
> Shorter. I suggest you start there. The sense of transparency AS
> transparency is to some extent in the ear of the beholder. What will
> be most affected is the timbre of instruments. The distortion spectrum
> as produced by many SETA's apparently makes instruments more palpable,
> immediate, or "there" for many listeners. Other listeners find that
> the presentation is due to artifacts.

Snip

> SETA's are no panacea. Even the best SETA can be shown to have
> problems, and virtually any listener can, with persistence, be trained
> to hear the artifacts AS artifacts rather than as "transparency". It
> just so happens that for many listeners the willing suspension of
> disbelief occurs more quickly and easily with a good SETA in the
> system. I can't argue with that.

It is worth noting though, the most enthusistic responses are
initially from novice (untrained) listeners. Are they more attuned to
'natural' sounds? Does experience and doctrine breed it out of us?

I wonder Scott, just how much of the poor traits of an SE amp are due
to a less than stellar power supply? Would a stable (highly regulated)
output supply be modulated less, and in the end offer a more stable
output? Is this partially responsible for the lower distortion in P-P
amps? Can they (P-P amps) also be viewed as offering, in practice,
active shunt regulation to an SE curcuit (up to the x-former)?

> They do--in SETA's. The SETA movement is to a large extent a movement
> away from accuracy as a criterion and toward beauty, emotion, and
> sheer involvement. It creates the strongest line of demarcation yet
> between tubes and transistors.

> Anything present at the output (other than gain) which is not present
> at the input, is distortion. Electrode resonances are signal-induced,
> caused by electrons impacting the anode plate and bouncing off and
> around inside the tube like BB's (thus demonstrating the particle
> nature of electrons!).

Or the diffracted light shining on your wall when you are listening to
your tube amp in the dark (demonstrating the wave dichotemy of
electro-magnetics in nature). It's all in how you *chose* to perceive
it. And work with it.

> The tube thus gets it from within and from without.

George Harrison? Sgt. Pepper's? ;)

> Harmonic distortion is a purely electrical effect due to curvature of
> the device conductance characteristics. IM distortion is due to sum
> and difference products between the spurious harmonic products. Noise
> can modulate these products, further altering the perceived
> result. Resonance adds spice to the mix. :-0

Or of of adding the missing trace elements to your
diet. Unfortuneately for me I've found the current state of store
bought home music reproduction the audio equivalent of "Solient Green"
:-(.

-Steve Jones

[Steve Jones, contractor]

Scott Frankland wrote:

> Well, I'm planning to test this theory by designing a very low
> distortion (i.e., low wideband distortion) tube amp. I intend to do it
> for both a pentode PP and an OTL SETA (triode).

Since pentodes (and beam power tubes) exhibit a proportionally larger
percentage of odd order harmonics than triodes, what will be your
*main* corrective action taken to remove them?

Triode strap the output tubes?

Ultra-linear operation with minimal local fedback?

Pentode operation with heavy global feedback?

If you run true triodes P-P, won't less 'corrective' action need to be
taken? With the outputs balanced, the predominate even harmonics would
be nulled and the remaining lower in level odd harmonics would be much
more easily 'corrected'.

Just a thought.

-Steve Jones

[sdura...@aol.com]

In article <reenh0k...@eyrie.graphics.cornell.edu>, Scott Frankland
<audi...@ix.netcom.com> writes:

>In previous posts I have referenced Olson, Winckel, Ladner, and
>Shorter. I suggest you start there. The sense of transparency AS
>transparency is to some extent in the ear of the beholder. What will
>be most affected is the timbre of instruments. The distortion spectrum
>as produced by many SETA's apparently makes instruments more palpable,
>immediate, or "there" for many listeners. Other listeners find that
>the presentation is due to artifacts.

Could you further explain this perception? What do you mean by artifacts?
If the "other" listeners also hear more palpability, immediacy or
"thereness," do these "artifacts" somehow prevent them from enjoying the
sound?

>> It seems to me, then, that there should be an ideal characteristic
>> and amount of bass distortion causing these midrange harmonics that
>> increase the subjective sense of transparency.
>
>That is, if you are willing to tolerate bass distortion, which I am not.

I'm not sure I understand. If the goal of audio is to provide as
transparent a window to the original event, would it not be preferable
to put up with some bass distortion to get a better sense of
transparency? It's the same sort of thing as with your reference to
"artifacts" above.

All amplifiers are compromises. It would seem to me that the very
best reproduction is the one that is most transparent, regardless of
specific parameters such as frequency response, distortion, dynamics,
etc. These are merely building blocks towards the realization of
transparent presentation. So if an amplifier provides a marvelous
sense of transparency, who cares if it doesn't have great bass, lacks
in dynamics, is sizzily in the highs or has a flat soundstage. The
sense of transparency that great amplifiers (and audio gear, in
general) have is really one of the whole being greater than the sum of
the parts.

>SETA's are no panacea. Even the best SETA can be shown to have
>problems, and virtually any listener can, with persistence, be trained
>to hear the artifacts AS artifacts rather than as "transparency". It
>just so happens that for many listeners the willing suspension of
>disbelief occurs more quickly and easily with a good SETA in the
>system. I can't argue with that.

I think that the quicker suspension of disbelief that occurs with SETAs
goes a long ways towards establishing the idea that "transparency" should
be the primary goal in audio. And I agree that people hear transparency
differently and that SETAs may not be the best solution for all.

>They do--in SETA's. The SETA movement is to a large extent a movement
>away from accuracy as a criterion and toward beauty, emotion, and
>sheer involvement.

I agree with this, except that "moving away from accuracy" is not
consistent with the greater sense of transparency that SETAs provide.
Transparency must be, by definition, more accurate. So what's at
issue is that SETAs are technically less accurate but perceptually
more transparent, which goes back to my whole sense that there's no
(or very poor) link between technical and perceptual accuracy,
published studies not withstanding.

>Well, I think it [distortion] should be eliminated, but that doesn't seem

to do it
>for me either, for reasons which I can't explain just yet. I think
>that, if there is in fact distortion present, it should be smoothly
>converging, and should contain even as well as odd products (for the
>reasons stated in a previous post).

I get a sense that there's an ideal model of distortion and dynamic
range components in a hi-fidelity audio signal that provides the best
sense of transparency to a wide range of listeners. The idea that
this model would be the lowest possible distortion and the widest
dynamic range makes sense on paper, but there are too many examples,
both anecdotal and tested, of situations where distortion and
compression lead to a better sense of capturing the original event. I
would envision that day when a simple hearing test at audio salons
could allow the listener to identify their hearing characteristics.
Manufacturers would offer gear that meets different distortion and
dynamic range criteria and listeners would be much better served,
rather than blindly groping about as they do now.

>It will be different for every amplifier. That much I can state
>unequivocally. I recommend that you make a study of the Stereophile
>back issues relating to amplifiers. Look at the distortion plots for
>distortion vs. load vs. power output vs. frequency, etc. You will find
>that it is not a simple question to answer.

No question. I can't even correlate the published plots versus the
reviewer's subjective comments about the component. The whole thing
is a total mystery to me.

If you were allowed to listen to two amplifiers, say a Spectral and a
Levinson, or an Audio Research and one of your tube amps, then were
handed two unidentified plots (showing the parameters you described
above), could you determine which plot belonged to which amplifier?

[Joe Berry]

sdura...@aol.com wrote:

>Are there any indications in the scientific literature that support
>the notion of added harmonics creating a sense of "transparency" (NOT
>euphonics or "pleasing", but transparent in the sense that the subject
>has a heightened aural sense of the original soundfield)?

I don't know about the scientific literature, but I am aware of
interesting parallels in the field of photography, where various
techniques are routinely used to enhance the "readability" of the
image, for both analytical and aesthetic purposes.

At least in that realm of endeavor, it's easy to see that "distortion"
is a somewhat relative concept. If the eye needs certain clues in
order to make sense of a 2D rendering of 3D (or 4D) reality, and if
those clues are enhanced such that the intelligiblity of the image is
improved, is the enhanced image a greater distortion of the reality it
represents than the original unadulterated image, or a lesser one?

--JB

[Steven Abrams]

sdura...@aol.com writes:
> While I prefer non-linear components in certain cases, I don't
> perceive them as being "euphonic." Let me explain why.

Oh, NOOOOOoooooo!!!! And we were doing so well for so long!

> The idea of euphonic or pleasing to the ear requires a value
> judgment on the aesthetic nature of what one is hearing. Proper

> audio reproduction should require no aesthetic judgment - it should

> merely be an assessment of how closely the reproduction matches the
> person's perception of what live music sounds like to them.

This paragraph has a serious logical flaw.

If I read right, you are opposed to the description of those audio
products which you prefer as euphonic, or pleasing to the ear. Your
reason is that this is an aesthetic judgment -- it sounds nicer or
more pleasing, rather than more live. "Sounding" more live is a value
judgment that each person makes for him or her self. Since we all
agree that live music is the most pleasing, wouldn't anything which
makes reproduced audio, for any given person, sound more live also
make it sound more pleasing to that person's ear? Wouldn't that be
euphonic?

> [...] But when trying

> to achieve a sound that is as close to live as possible, I've found that
> certain non-linear products get me closer to that sense of live sound.

You used the word "sense" -- which is good. It sounds, to you, more
like live sound. Therefore it sounds *better* to you. Therefore, it
is euphonic. This is not a bad word, nor a criticism. It's just what
the word means. "Sweet-voiced;" Musical; Pleasing to the ear.

> I understand that our ears are not the same and that there may be some out
> there who find that the MOST linear products best replicate live sound.
> But if you take into consideration how we hear differently, I can't see
> how SO many people naturally fall into the category of preferring linear
> products.

Now you have apparently decided to attack with the same old claim that
you've been dragging up, and I'm frankly sick and tired of it and I
really wish you'd shut up with it already because it's quite annoying.
The claim made by you is that the only reason so many people disagree
with you is because they haven't *listened*. Be serious. Nobody
hangs out on a high-end newsgroup who doesn't listen to the music.

> Many of these people really haven't taken the time to fully
> explore their hearing response and matching it to the best audio products
> for their hearing. They merely accept linear is correct because it is
> mathematically proper and let it go at that.

MANY? I don't know where you get this from.

Let's talk about many.

Garbage systems FAR OUTSELL quality systems. If you take all of the
audio systems sold in the world, what percentage of them do you think
consist of equipment that *ever* gets mentioned favorably in
Stereophile, TAS, and all of the other high-end rags combined? Let's
focus in on the speakers. Ever listen to those department-store rack
systems? How about the Circuit City specials? The speakers are
typically so awful that neither the worlds most linear solid state amp
nor the most romantic, warm, tube system could possibly handle it.
The amps *might* measure linearly if you measured them driving a nice
stable 8 ohm resistor, driven nowhere near clipping, and viewed your
scope from a great distance.

If you're counting every owner of solid-state equipment in the group
of people who buy linear systems because they accept, blindly, that it
is correct, you're missing the point totally. Buyers of these
department-store systems typically don't care about sound quality,
don't know how to judge sound quality, or do care but don't realize
that good quality is affordable if you're careful. So in your
assertion of "many" we can't count these people. Even if they think
that they're buying an "accurate" system because of some marketing
specs thrown on the front of the pressboard rack, they aren't. Watch
your step: I think the arms of your strawman just fell off.

So now you're left talking about people who purposely shop for quality
equipment and end up with something ranging from NAD, Rotel, Adcom,
Meridian, all the way up to Krell and Levinson. Do you really think
that these people don't listen to the music? Do you really think that
they blindly pick amongst those contenders for whichever has the
"flattest" frequency response? If this is what you think, you've got
no idea what you're talking about. I certainly spend my fair share of
time listening to music when I shop for equipment, and I've got my
fair share of experience listening to live music, and I pick whatever
gets me *there* the best. Careful you don't trip over your strawman's
legs.

Now, think about what you said above: You asked how is it, given that
we all hear differently, that SO many people prefer linear products.

I'll combine this with your previous statements that there is an
overwhelming resurgence in SETAs and LPs, taking the world by storm.
Why don't you find yourself wondering why SO many people prefer LPs
and SETA, asking if instead of *listening* and picking whatever gets
them closest to the sense of live music, they are just following the
trend? (Not that this is what I think is happening -- I just thought
it was interesting to note the type of spin you're putting on things.)

> Are you one of these people? Have you taken the time to test
> yourself and truly MAKE SURE that linear products offer the closest
> sense to your perception of live music?

Face facts: SOME people prefer vanilla. SOME people prefer
chocolate. Given that different people have different sets of
taste-buds and different ways of perceiving flavors, are you surprised
that SO many people like chocolate?

It ain't no big deal. Get over it, and stop trying to PROVE that
you're preference is right.

~~~Steve
(PS: There's a pile of hay that might have been a torso. Please take
it with you when you go.)
--
Steven Abrams abr...@cs.columbia.edu

Living is easy with eyes closed, misunderstanding all you see.
-Lennon/McCartney

[Joe Berry]

sdura...@aol.com wrote:

>Are there any indications in the scientific literature that support
>the notion of added harmonics creating a sense of "transparency" (NOT
>euphonics or "pleasing", but transparent in the sense that the subject
>has a heightened aural sense of the original soundfield)?

I think it's pretty well established that human hearing relies on
second harmonic distortion for information about sound level and
dynamics. If that's true, enhancing the second harmonic content of a
recording should allow a clearer "sense of the original soundfield"
insofar as that sense is derived from volume and dynamic clues.

It seems plausible to me that, if the added distortion is similar in
amount to that found in nature and correlates with the input signal
(is in phase with it and rises and falls in proportion to it), it will
not be heard as coloration, but as an integral part of the signal.

Whle adding distortion in this way would not increase the true
information content of the recording, it might very well increase the
emotional "intelligibility" of the existing content.

[sdura...@aol.com]

In article <58k46t$e...@canyon.sr.hp.com>, Scott Frankland
<audi...@ix.netcom.com> writes:

>Your definition of "proper audio reproduction" is interesting, but it
>fails to distinguish between audio criticism and the pure listening
>experience, which I feel is an important distinction. When we listen
>to music played back, we do not make "assessments" unless we are in
>the critical mode. The only way to become involved with the music is
>to drop the critical mode, i.e., to willingly suspend our disbelief.
>In so doing, we enter the realm of pure aesthetic experience. These
>two modes, therefore, are to a large extent self-defeating.

Self-defeating? Mutually exclusive, maybe, but I'm troubled by your
statement of self-defeating.

Personally, I found that the two modes are really stages. The first
stage is to evaluate a component or system by critically evaluating
the quality of audio playback on its own. Once certain standards have
been met, I can then segue into the "music-listening" mode, which I
agree is purely aesthetically-based. And, most importantly, I've
often found that certain sonic anomalities come up during this later
stage that are often in contradiction with the primary, "audio-only"
assessment. This has been notably true in digital audio, whereby the
audio quality on its own is very good, but after listening to music,
certain subliminal effects detract from enjoying the music as much as
one would listening to analog.

>Whether the music sounds live or not is a separate issue. Live music
>is the standard because it sounds *better* than reproduced music. This
>is not to say that it might not be possible to someday reverse these
>roles. The problem here is akin to building a better sounding violin
>than a Stradivarius.

I don't entirely agree. Live music is a reference. It may, or may
not, sound good or pleasing. It should sound "correct," though. How
do we define "correct." To each individual, it is what they hear when
they are at a live venue. Now it is up to each individual to remember
that sound and try to associate recorded sound with that memory.
Potential for inconsistency? Huge. Is that a problem? Many
scientists seem to think so, but, in my view, we each can PERSONALLY
make aour best effort to become as aware as possible of the
characteristics of live sound and concentrate like the dickens when
listening to reproduced music to evaluate the sound. Some of us get
close, others not. Hence, all our gear sounds different and we argue
with each other about what we're hearing. I'm afraid that's the best
we can do, for now.

Siegfried

[.sig deleted -- bt]

[jj, curmudgeon and all-around grouch]

In article <58k4su$e...@canyon.sr.hp.com> sdura...@aol.com writes:

>>The distortion spectrum as produced by many SETA's apparently makes
>>instruments more palpable, immediate, or "there" for many listeners.
>>Other listeners find that the presentation is due to artifacts.

>Could you further explain this perception? What do you mean by
>artifacts? If the "other" listeners also hear more palpability,
>immediacy or "thereness," do these "artifacts" somehow prevent them
>from enjoying the sound?

The difficulty is that listeners have many levels of experience and
that different people, for phisiological reasons, have different
sensitivities.

For instance, many of the distortions that sounded "nice" or
"euphonic" or "more real" to me 20 years ago are now annoying as I've
taken lots of time to learn the results of individual and combined
distortions. (Working on perceptual coding for 12 or 13 years is
partially a study in every form of distortion known to mankind,
including some that should have never BEEN known :-)

Basically, what a listener prefers today may change with training and
experience, time of day, listening paradigm, and so on. So, what is
sauce for one is sewage for another. Really. That's how it works out.

>I'm not sure I understand. If the goal of audio is to provide as
>transparent a window to the original event, would it not be preferable
>to put up with some bass distortion to get a better sense of
>transparency?

Not if that bass distortion that you've learned to hear interferes
with the "reality" more than it provides euphony. The more trained
you are in distortion, very often the more you dislike what others
prefer as euphonic. (but not ALWAYS, please!)

>I think that the quicker suspension of disbelief that occurs with SETAs
>goes a long ways towards establishing the idea that "transparency" should
>be the primary goal in audio. And I agree that people hear transparency
>differently and that SETAs may not be the best solution for all.

I find an SETA uncomfortable to listen to, and the result nothing near
either transparent, suspension of belief, or anything of the sort. In
short, the ones that are commonly put forth sound BAD to me.
Sorry. Preference is preference. Prefer what YOU prefer, but don't
judge or predict other's preference.

>I get a sense that there's an ideal model of distortion and dynamic
>range components in a hi-fidelity audio signal that provides the best
>sense of transparency to a wide range of listeners.

Your 'sense' is not yet born out by experience, sorry to say.
Different preferences prefer different distortions.

>No question. I can't even correlate the published plots versus the
>reviewer's subjective comments about the component. The whole thing
>is a total mystery to me.

Most of the published plots are not particularly meaningful, so that's
not perhaps too surprising.

--
Copyright alice!jj 1996, all rights reserved, except transmission by USENET
and like facilities granted. This notice must be included. Any use by a
provider charging in any way for the IP represented in and by this article
and any inclusion in print or other media are specifically prohibited.

[Scott Frankland]

Joe Berry wrote:

> I don't know about the scientific literature, but I am aware of
> interesting parallels in the field of photography, where various
> techniques are routinely used to enhance the "readability" of the
> image, for both analytical and aesthetic purposes.
>
> At least in that realm of endeavor, it's easy to see that "distortion"
> is a somewhat relative concept. If the eye needs certain clues in
> order to make sense of a 2D rendering of 3D (or 4D) reality, and if
> those clues are enhanced such that the intelligiblity of the image is
> improved, is the enhanced image a greater distortion of the reality it
> represents than the original unadulterated image, or a lesser one?

Guess it depends on how hard you concentrate. If you look at those
funny computer-enhanced stick people in the newer kiddy films with a
subdued eye, they can almost come to life! OTOH, some of these
photo-shopped pin-up girls (that look too-too perfect) are hard to
criticize unless you really concentrate. Enhancement has its
place. ;-)

~SF~

[Scott Frankland]

Steve Jones, contractor wrote:

> It is worth noting though, the most enthusistic responses are
> initially from novice (untrained) listeners. Are they more attuned to
> 'natural' sounds? Does experience and doctrine breed it out of us?

As we learn to detect distortion AS distortion, regardless of whether
it is euphonic or not, I find that we tend to gravitate toward an
absence of ALL distortion. This might be called the PURE definition of
transparency.

> I wonder Scott, just how much of the poor traits of an SE amp are due
> to a less than stellar power supply? Would a stable (highly regulated)
> output supply be modulated less, and in the end offer a more stable
> output?

It would certainly help with bass control. Power supply damping is an
integral element of overall damping. It might also reduce treble
distortion, enhance definition, and improve transient response. OTOH,
a poorly executed regulated supply might KILL the sound.

> Is this partially responsible for the lower distortion in P-P
> amps? Can they (P-P amps) also be viewed as offering, in practice,
> active shunt regulation to an SE curcuit (up to the x-former)?

Well, there is certainly more power supply rejection in PP. The power
supply is quite critical in either case, IMO.

Regards,

~SF~

[Sheldon D. Stokes]

In article <58hv3v$j...@tolstoy.lerc.nasa.gov>, "Steve Jones, contractor"
<sjo...@vcd.hp.com> wrote:

> I wonder Scott, just how much of the poor traits of an SE amp are due
> to a less than stellar power supply? Would a stable (highly regulated)
> output supply be modulated less, and in the end offer a more stable

> output? Is this partially responsible for the lower distortion in P-P

> amps? Can they (P-P amps) also be viewed as offering, in practice,
> active shunt regulation to an SE curcuit (up to the x-former)?

I'd look at the output transformer. You are fighting the laws of
physics with SE output transformers. A good transformer should have
very strong magnetic coupling between the primary and secondary (ref.
radiotron). But with very good primary magnetic coupling to the core
(and thus to the secondary), you run into core saturation problems
when running DC through the primary as you have to do with all single
ended amps. So you have to not couple the primary to the core so
closely to aviod core saturation. The way this is done is to air gap
the core. But not the primary is no longer closely coupled to the
secondary. And now your hysterisis goes through the roof and the
frequency response of the transformer gets much worse. You can
improve the frequency response by makign the transformer much larger,
but then you run into stray capacitance problems that will limit your
high frequency response.

There is a reason that SE tube amps sound crappy.

If you want to try single ended, I'd recommend that you build the PASS
Zen, it's single ended MOSFET. MOSFETs are also transconductance
devices much like tubes.

With push pull output transformers, the DC curent is introduced in the
center of the primary winding and travels to both ends, thus canceling
out the net magnetic field. Also the power supply ripple is neatly
cancelled out with a push-pull output.

Sheldon

[Scott Frankland]

sdura...@aol.com wrote:
>
> Could you further explain this perception? What do you mean by artifacts?

Something left over in the process of baking the cake. Distortion products.

> If the "other" listeners..
[snip -- bt]

I can't speak for other listeners.

> >> It seems to me, then, that there should be an ideal characteristic
> >> and amount of bass distortion causing these midrange harmonics that
> >> increase the subjective sense of transparency.
> >
> >That is, if you are willing to tolerate bass distortion, which I am not.
>
> I'm not sure I understand.

[snip --bt]
> ...transparent a window to the original event, would it not be preferable

> to put up with some bass distortion to get a better sense of
> transparency?

[snip -- bt]

Well, why not have it all?

> >They do--in SETA's. The SETA movement is to a large extent a movement
> >away from accuracy as a criterion and toward beauty, emotion, and
> >sheer involvement.
>
> I agree with this, except that "moving away from accuracy" is not
> consistent with the greater sense of transparency that SETAs provide.
> Transparency must be, by definition, more accurate.

Not more accurate; more apparently real to a particular listener.

> So what's at
> issue is that SETAs are technically less accurate but perceptually
> more transparent, which goes back to my whole sense that there's no
> (or very poor) link between technical and perceptual accuracy,
> published studies not withstanding.

You're not exactly encouraging the process, here, Seigfreid.

[snip -- bt]

> ...there are too many examples,

> both anecdotal and tested, of situations where distortion and
> compression lead to a better sense of capturing the original event.

Why not just make better recordings, microphones and speakers? Then
we could all move toward pure transparency, i.e., an absense of all
distortion products.

> I would envision that day when a simple hearing test at audio salons
> could allow the listener to identify their hearing characteristics.
> Manufacturers would offer gear that meets different distortion and
> dynamic range criteria and listeners would be much better served,
> rather than blindly groping about as they do now.

What blindly groping? The first step is to listen to a good solid
state amp. Next listen to a good SETA amp. This will give the widest
divergence of sound qualities among amplifiers. If they each now sound
too extreme, try a PP tube amp, which is somewhat in the middle. We
now have chocolate, vanilla, and strawberry-- something for
everyone. If you want mint swirls in your chocolate, we got that, too.

> No question. I can't even correlate the published plots versus the
> reviewer's subjective comments about the component. The whole thing
> is a total mystery to me.

Which is why I took the trouble to write an article for Stereophile.
There ARE correlations. One just needs to learn how to tune into them.

> If you were allowed to listen to two amplifiers, say a Spectral and a
> Levinson, or an Audio Research and one of your tube amps, then were
> handed two unidentified plots (showing the parameters you described
> above), could you determine which plot belonged to which amplifier?

If they sound different, Siegfried, then they will measure different.
Finding correlations is an essential part of engineering and design.
Please don't limit the possibilities with unnecessary skepticism.

~SF~

[dpl...@iq.navio.com]

In article <58k4su$e...@canyon.sr.hp.com> sduraybito writes:

>Could you further explain this perception? What do you mean by artifacts?

> If the "other" listeners also hear more palpability, immediacy or
>"thereness," do these "artifacts" somehow prevent them from enjoying the
>sound?

In some listeners, this is apparently the case.

>>> It seems to me, then, that there should be an ideal characteristic
>>> and amount of bass distortion causing these midrange harmonics that
>>> increase the subjective sense of transparency.
>>
>>That is, if you are willing to tolerate bass distortion, which I am not.
>

>I'm not sure I understand. If the goal of audio is to provide as

>transparent a window to the original event, would it not be preferable
>to put up with some bass distortion to get a better sense of
>transparency?

Preferable TO WHOM, please?

Once again, you seem to be falling into the trap of believing that
there is necessarily one Ideal form of audio presentation - an
"optimum" which is necessarily the goal for all listeners.

I believe this is a false goal. People differ in their perceptions.
As JJ (I believe) mentioned some time ago, different people have been
shown to process sound differently in their ear/brain systems, and to
respond to different cues in the sound. I suspect (based on various
things I've read over the past few years) that this is probably due to
a combination of factors... actual "wiring differences" in the nervous
system (inborn, and possibly genetically-influenced), the native
language of the listener (PET-scan research indicates that the brains
of first-language speakers of Japanese process language and sound
differently than the brains of first-language English speakers), and
personal environment and experience.

It's entirely possible that one person may achieve the most pleasing
subjective experience with a setup that maximizes "immediacy", and not
be bothered by (or perhaps even notice) a higher level of bass
distortion, or a rolloff in the low bass or upper treble. Another
person might have a very different experience, finding the distortion
to be offensive enough to interfere with the listening experience and
thus detract from the "immediacy" or "transparency".

In this case, whose listening experience is valid? Whose preferred
mix of linearities, nonlinearities, and distortions is correct?

To my mind, the only answer is "Both are!" To say otherwise is to
force people into a Procrustian bed, to no purpose other than the
personal goal of being "right". Bah.

> It's the same sort of thing as with your reference to
>"artifacts" above.
>
>All amplifiers are compromises. It would seem to me that the very
>best reproduction is the one that is most transparent, regardless of
>specific parameters such as frequency response, distortion, dynamics,
>etc. These are merely building blocks towards the realization of
>transparent presentation.

Once again - transparent _to_ _whom_? You are attempting to elevate
"transparency" to an absolute. I don't believe it is, or can be.

"Transparency" is a subjective reaction - it's a resultant, not a
fundamental entitity in and of itself. You can't leave the listener
out of the equation!

> So if an amplifier provides a marvelous
>sense of transparency, who cares if it doesn't have great bass, lacks
>in dynamics, is sizzily in the highs or has a flat soundstage.

Who cares? Scott cares, obviously! TO HIM, some of these defects are
INCOMPATIBLE with a "marvelous sense of transparency". TO HIM, a
better sense of transparency/immediacy requires that these defects not
be present, because (to his nervous system and listening habits) these
defects are offensive enough to murk up the listening experience. Or
so I read his words...

> The
>sense of transparency that great amplifiers (and audio gear, in
>general) have is really one of the whole being greater than the sum of
>the parts.

But, it's not an absolute. Based on your comments and Scott's, it's
apparent that two different people may rank the relative
"transparency" of different components in different orders.

>I think that the quicker suspension of disbelief that occurs with SETAs
>goes a long ways towards establishing the idea that "transparency" should
>be the primary goal in audio. And I agree that people hear transparency
>differently and that SETAs may not be the best solution for all.

... but I don't think you necessarily realize the _amount_ of
variability in what different people will find contributes to their
own ability to sense "transparency".

>>They do--in SETA's. The SETA movement is to a large extent a movement
>>away from accuracy as a criterion and toward beauty, emotion, and
>>sheer involvement.
>

>I agree with this, except that "moving away from accuracy" is not
>consistent with the greater sense of transparency that SETAs provide.
>Transparency must be, by definition, more accurate.

I disagree. You are comparing apples and oranges here.

> So what's at
>issue is that SETAs are technically less accurate but perceptually
>more transparent, which goes back to my whole sense that there's no
>(or very poor) link between technical and perceptual accuracy,
>published studies not withstanding.

#sigh#. Still resisting the idea that certain distortions can be
euphonic, I guess.

I suggest the following: an increased sense of transparency and
immediacy _is_ euphonic. You like it, you consider it to be an
ultimate goal... clearly, it's euphonic (literally, "good sounding" I
believe).

I also suggest that some of the well-characterized euphonic
distortions and nonlinearities (frequency-response irregularities,
even-order harmonic distortion, dynamic effects, microphonics) may be
responsible for at least part of what's contributing to the
sense-of-transparency that you favor.

[Pete Goudreau]

> On Fri, 06 Dec 1996, Scott Frankland <audi...@ix.netcom.com> wrote,

> > Steve Jones, contractor wrote:

> <snip>

> > -My next amp is a SET (in the works). BTW, Scott, any (re)sources for
> > information on high frequency filament supplies for a DHT (5VAC @
> > 1.2A, ~170khz -solid state will do)?

> I'd be willing to bet that Pete Goudrea can answer this
> question. Pete?

Sure thing Scott and Steve...

Although I can't fathom the desire to power the filament of a vacuum
tube with a high freq AC waveform, it is doable. Did it with
switchers for plasma ion sources and it works fine albeit with serious
EMI generation. If the source was filtered for sinusoidal output or
set up as a resonant converter than this problem would be minimized
but 170KHz energy would be coupled across the heater to cathode and
heater to grid capacitances and be present in the plate circuit where
it could generate intermod products...not a good thing.

Personally, I would recommend the use of DC for filaments in audio
tube circuits with extensive differential and common mode filtering
and shielding as this is a great place for RF to enter the circuit and
wreak havoc. Adding harmonic dampers (basically discrete equivalents
of ferrite beads) at the grids are also pretty handy to minimize RF
ingress in the stage as well.

As to using high freq AC on the filaments...I suggest that the
designer avoid this like the plague and concentrate on DC sources and
appopriate filtering instead...just my $0.02, thanks for asking
though...

Cheers,
Pete

[Scott Frankland]

Steve Jones, contractor wrote:
>
> Since pentodes (and beam power tubes) exhibit a proportionally larger
> percentage of odd order harmonics than triodes, what will be your
> *main* corrective action taken to remove them?
> Triode strap the output tubes?

Then it wouldn't be pentode.

> Ultra-linear operation with minimal local fedback?

Something like that. Ultralinear gives a maximum feedback of roughly
6db. With cathode feedback, such as Walker used in the Quad amps,
additional local feedback can be used. I plan to use considerably more
local feedback than Walker used. The only limitation on this type of
feedback is the peak driver swing required to overcome the high bias
voltage that appears at the pentode control grids due to the decrease
in input sensitivity that accompanies local feedback.

Rather than use McIntosh's solution of the bootstrapped driver with
positive feedback, I plan to design a driver that can swing something
like 150V peak without the bootstrap. The bootstrap is partly
responsible for the increase in midrange IM products that occurs in
the Mac due to the positive feedback.

> Pentode operation with heavy global feedback?

This is an interesting approach that has been endorsed by such
theorists as Edward Cherry and Norman Crowhurst. See, e.g., Cherry and
Dabke, "Transient Intermodulation Distortion--Part 2: Soft
Nonlinearity (JAES, 1986 Jan/Feb). The authors explain how negative
feedback, when applied around a square-law nonlinearity like a triode,
creates a converging series of higher order distortion products; but
when properly applied around a cube-law nonlinearity like a pentode,
the opposite occurs--i.e., the higher harmonics that were there in the
absence of feedback are reduced by the feedback.

Crowhurst goes into more detail in his excellent "The Amplifier
Distortion Story" (Audio, 1959 Apr/May). Later, he cautions against
the use of loop feedback in pentodes with reactive loads in his
"Feedback--Head Cook and Bottle Washer!" (Audio, 1962 Jan). This is
the same problem that Blumlein solved with his screen-feedback
transformer connection in the '30s. Walker then took Blumlein's idea
and turned it upside down! How soon we forget the lessons of history.

> If you run true triodes P-P, won't less 'corrective' action need to be
> taken? With the outputs balanced, the predominate even harmonics would
> be nulled and the remaining lower in level odd harmonics would be much
> more easily 'corrected'.

What about damping? What about treble distortion? These will suffer.
Feedback, properly applied, corrects all of these problems
simultaneously. Cherry's final solution to these problems involves
the use of multiple nested loops. I plan to use just two loops--one
global, and one local to the output stage (the major distortion
producing stage). Pentodes are used partly for the reason stated
above, and partly because they have more gain than triodes and
therefore more local feedback is available with pentodes.

~SF~

[Steve Jones, contractor]

Scott Frankland wrote:

> What blindly groping? The first step is to listen to a good solid
> state amp. Next listen to a good SETA amp. This will give the widest
> divergence of sound qualities among amplifiers. If they each now sound
> too extreme, try a PP tube amp, which is somewhat in the middle. We
> now have chocolate, vanilla, and strawberry-- something for
> everyone. If you want mint swirls in your chocolate, we got that, too.

Good analogy, Scott (the air brushed pin-up girl was better though
:-).

And here's a good time for me to take a breather.

Much appreciated your answering of my questions on this thread. Gave
me alot of understanding of your priorities in the juggling of the
many variable involved in making music.

But there are as many different mind sets on this newsgroup as there
are contributors. One size fit all? Nope. You've given me some new
insights. But, for me I still prefer SE.

To put this in a less serious vein (Hi-Fi is SO serious now), alot of
the SE proponents seem to be on a missionary zeal to reform the church
and convert the pagan. Ain't gonna happen. This Martin Luther (of
audio) Reformation will not bring the establishment to it's knees.

And Siggy's no John Knox.

Enjoy the music,
-Steve Jones

[Sheldon D. Stokes]

> If you run true triodes P-P, won't less 'corrective' action need to be
> taken? With the outputs balanced, the predominate even harmonics would
> be nulled and the remaining lower in level odd harmonics would be much
> more easily 'corrected'.

right, but you'll have the "miller effect" to deal with and then
you'll have to build much more robust driver stages for the finals,
which will cause more distortion. It's a trade off, there is no magic
pill.

The real evil element in ANY tube amp is those output transformers.
The best transformers ever made were bad, and then they go down from
there. Take a look at how bad a transformer butchers a signal, it's
pretty scarey when you think about it.

I still use tube amps and probably will until there are no more tubes,
but I also know the limitations of the design type.

Sheldon

[Scott Frankland]

sdura...@aol.com wrote:

> In article <58k46t$e...@canyon.sr.hp.com>, Scott Frankland
> <audi...@ix.netcom.com> writes:

> Self-defeating? Mutually exclusive, maybe, but I'm troubled by your
> statement of self-defeating.

Mutually exclusive, antithetical, or self-defeating if pursued
simultaneously.

> Personally, I found that the two modes are really stages. The first
> stage is to evaluate a component or system by critically evaluating
> the quality of audio playback on its own. Once certain standards have
> been met, I can then segue into the "music-listening" mode, which I
> agree is purely aesthetically-based.

Martin DeWulf calls it "crossing the threshold". Others have called
it the willing suspension of disbelief.

>> Whether the music sounds live or not is a separate issue. Live
>> music is the standard because it sounds *better* than reproduced
>> music. This is not to say that it might not be possible to someday
>> reverse these roles. The problem here is akin to building a better
>> sounding violin than a Stradivarius.

> I don't entirely agree. Live music is a reference. It may, or may
> not, sound good or pleasing. It should sound "correct," though. How
> do we define "correct." To each individual, it is what they hear when
> they are at a live venue. Now it is up to each individual to remember
> that sound and try to associate recorded sound with that memory.

This is an analytical error. Memory is not satisfied by the music; it
is *human sensibility* that is satisfied. Our MEMORY of live music is
what PREVENTS US FROM BELIEVING that a reproduced sound is live. Our
sensibility is thus modified by the "tyranny" of memory. Memory is
what prevents us from enjoying the experience as a pure aesthetic
experience. In common parlance, this is known as being spoiled.

Going a little further...does reproduced music have to convince us
that it's "correct"; i.e., that it "matches" our memory? No. It merely
has to sound AS GOOD AS live; in some way, shape, or form. In other
words, none of the nuance of live sound must be missed due to the
tyranny of memory. This, I contend, may one day be achieved via
digital enhancement (much as film is now enhanced by digital
enhancement). The promise of digital is high speed--the ability to
enter into a time-sensitive program and modify it in real time. Audio
signals virtually stand still compared to what is now possible in the
digital realm.

> Potential for inconsistency? Huge. Is that a problem? Many
> scientists seem to think so, but, in my view, we each can PERSONALLY
> make aour best effort to become as aware as possible of the
> characteristics of live sound and concentrate like the dickens when
> listening to reproduced music to evaluate the sound.

Well, be sure to come up for air once in awhile. There's a world of
music out there waiting to be discovered (and enjoyed).

~SF~

[sdura...@aol.com]

In article <58ptel$c...@canyon.sr.hp.com>, dpl...@iq.navio.com writes:

>In this case, whose listening experience is valid? Whose preferred
>mix of linearities, nonlinearities, and distortions is correct?
>
>To my mind, the only answer is "Both are!" To say otherwise is to
>force people into a Procrustian bed, to no purpose other than the
>personal goal of being "right". Bah.

What you say makes a lot of sense, yet I can't help thinking that
there's some sort of universal truth to all of this. It's akin to
sounds that comfort humans regardless of culture. You play Mozart and
there tends to be a common response by people all over the world. Why
would this also not hold true for a sense of proper music
reproduction?

>>I agree with this, except that "moving away from accuracy" is not
>>consistent with the greater sense of transparency that SETAs provide.
>>Transparency must be, by definition, more accurate.
>
>I disagree. You are comparing apples and oranges here.

I've asked Scott the same question: To a given listener, why are
transparency and accuracy not consonant? Under what conditions would
these two perceptions diverge? What kind of perception is
"inaccurate" transparency? Wouldn't it no longer be transparent? Is
there a condition of accuracy that does not also convey transparency?

>I suggest the following: an increased sense of transparency and
>immediacy _is_ euphonic. You like it, you consider it to be an
>ultimate goal... clearly, it's euphonic (literally, "good sounding" I
>believe).

In an absolute sense, transparency and euphonics do not necessarily go
hand-in-hand. Clearly one can have transparent sound of something
that is not good-sounding (i.e. a recording of something that sounds
ugly) and it is no longer euphonic. Further, I have found that there
exists certain audio gear that makes things sound too good - overly
warm, recessed highs, blurred micro-detail, etc - that convey a sense
of euphony but is not transparent. Surely, you must have similar
experiences.

>I also suggest that some of the well-characterized euphonic
>distortions and nonlinearities (frequency-response irregularities,
>even-order harmonic distortion, dynamic effects, microphonics) may be
>responsible for at least part of what's contributing to the
>sense-of-transparency that you favor.

I don't dispute this at all. I have found, though, that a certain mix
of distortions and non-linearities best suit my sense of transparency.
Other mixes don't do it. Now I'm not arguing that my mix is best for
everyone - it's just that it seems there's a considerable number of
people out here who like different mixes. Which still makes me wonder
what all this obsession with linear-ueber-alles is all about.

Siegfried

[sdura...@aol.com]

In article <58pta8$c...@canyon.sr.hp.com>, Scott Frankland
<audi...@ix.netcom.com> writes:

>> ...transparent a window to the original event, would it not be

preferable
>> to put up with some bass distortion to get a better sense of
>> transparency?

>[snip -- bt]
>
>Well, why not have it all?

I wish that were possible. To my ears, though, as you reduce
distortion and flatten frequency response (as with sold-state
amplifiers), the sound becomes flatter, less three-dimensional and,
overall, less involving. Other SETA listeners report the same
results.

>> >They do--in SETA's. The SETA movement is to a large extent a movement
>> >away from accuracy as a criterion and toward beauty, emotion, and
>> >sheer involvement.
>>

>> I agree with this, except that "moving away from accuracy" is not
>> consistent with the greater sense of transparency that SETAs provide.
>> Transparency must be, by definition, more accurate.
>

>Not more accurate; more apparently real to a particular listener.

To a particular listener, wouldn't a sense of "realness" be consonant
with being accurate? Under what conditions could these two
perceptions diverge?

>> So what's at
>> issue is that SETAs are technically less accurate but perceptually
>> more transparent, which goes back to my whole sense that there's no
>> (or very poor) link between technical and perceptual accuracy,
>> published studies not withstanding.
>

>You're not exactly encouraging the process, here, Seigfreid.

Sorry, it's just my opinion and I know it's not very popular. No
offense, please.

>> ...there are too many examples,
>> both anecdotal and tested, of situations where distortion and
>> compression lead to a better sense of capturing the original event.
>
>Why not just make better recordings, microphones and speakers? Then
>we could all move toward pure transparency, i.e., an absense of all
>distortion products.

Again, refer to my comment above wherein SETA listeners have
(apparently) come to dislike the sound of low distortion. How could
this be?

>> I would envision that day when a simple hearing test at audio salons
>> could allow the listener to identify their hearing characteristics.
>> Manufacturers would offer gear that meets different distortion and
>> dynamic range criteria and listeners would be much better served,
>> rather than blindly groping about as they do now.
>

>What blindly groping? The first step is to listen to a good solid
>state amp. Next listen to a good SETA amp. This will give the widest
>divergence of sound qualities among amplifiers. If they each now sound
>too extreme, try a PP tube amp, which is somewhat in the middle. We
>now have chocolate, vanilla, and strawberry-- something for
>everyone. If you want mint swirls in your chocolate, we got that, too.

If you look at the track record of the typical audiophile, you'll see
frequent equipment changes in search of the Holy Grail. Though a lot
of it is due to changing physiology and different expectations, a lot
of this behavior is simply a learning curve to match the right
combination of distortion and dynamic range to the listener's sense of
live music. From a consumer perspective, I perceive this as blindly
groping.

>> If you were allowed to listen to two amplifiers, say a Spectral and a
>> Levinson, or an Audio Research and one of your tube amps, then were
>> handed two unidentified plots (showing the parameters you described
>> above), could you determine which plot belonged to which amplifier?
>
>If they sound different, Siegfried, then they will measure different.
>Finding correlations is an essential part of engineering and design.
>Please don't limit the possibilities with unnecessary skepticism.

Mmmm. You didn't answer the question. I don't mean to put you on the
spot, Scott, but I would suspect that identifying an amplifier, blind,
based on its performance plots is no easier than identifying the type
of music on a CD by looking at it.

Siegfried

[Steve Jones, contractor]

Pete Goudreau wrote:

> Although I can't fathom the desire to power the filament of a vacuum
> tube with a high freq AC waveform, it is doable.

In a directly heated triode (300B), there is a difference in potential
from one end of the filament (+) to the other (-) if heated with DC. As
the filament is shaped as an inverted W, there will be physical varying
of cathode potential relative to the plate. AC (60hz) averages this
effect, with an attendant increase in hum.
In the RDH4, they mention 170khz heating as a way to move this
*possible* problem effectively out of band (audio). Just curious to try
it, thats all.

> If the source was filtered for sinusoidal output or
> set up as a resonant converter than this problem would be minimized
> but 170KHz energy would be coupled across the heater to cathode and
> heater to grid capacitances and be present in the plate circuit where
> it could generate intermod products...not a good thing.

In a 300B the filament is the cathode, and the inductance of the
output transformer *may* help dampen the RFI from the PS. Don't know
until I try it though.

> As to using high freq AC on the filaments...I suggest that the
> designer avoid this like the plague and concentrate on DC sources and
> appopriate filtering instead...just my $0.02, thanks for asking
> though...

Thanks for the time and your answers,

Typhoid Mary (oops!-Steve Jones)

[sdura...@aol.com]

In article <58ptel$c...@canyon.sr.hp.com>, dpl...@iq.navio.com writes:

>In this case, whose listening experience is valid? Whose preferred
>mix of linearities, nonlinearities, and distortions is correct?
>
>To my mind, the only answer is "Both are!" To say otherwise is to
>force people into a Procrustian bed, to no purpose other than the
>personal goal of being "right". Bah.

What you say makes a lot of sense, yet I can't help thinking that

there's some sort of universal truth to all of this. It's akin to
sounds that comfort humans regardless of culture. You play Mozart and
there tends to be a common response by people all over the world. Why
would this also not hold true for a sense of proper music
reproduction?

>>I agree with this, except that "moving away from accuracy" is not

>>consistent with the greater sense of transparency that SETAs provide.
>>Transparency must be, by definition, more accurate.
>
>I disagree. You are comparing apples and oranges here.

I've asked Scott the same question: To a given listener, why are

transparency and accuracy not consonant? Under what conditions would
these two perceptions diverge? What kind of perception is
"inaccurate" transparency? Wouldn't it no longer be transparent? Is
there a condition of accuracy that does not also convey transparency?

>I suggest the following: an increased sense of transparency and

>immediacy _is_ euphonic. You like it, you consider it to be an
>ultimate goal... clearly, it's euphonic (literally, "good sounding" I
>believe).

In an absolute sense, transparency and euphonics do not necessarily go

hand-in-hand. Clearly one can have transparent sound of something
that is not good-sounding (i.e. a recording of something that sounds
ugly) and it is no longer euphonic. Further, I have found that there
exists certain audio gear that makes things sound too good - overly
warm, recessed highs, blurred micro-detail, etc - that convey a sense
of euphony but is not transparent. Surely, you must have similar
experiences.

>I also suggest that some of the well-characterized euphonic

>distortions and nonlinearities (frequency-response irregularities,
>even-order harmonic distortion, dynamic effects, microphonics) may be
>responsible for at least part of what's contributing to the
>sense-of-transparency that you favor.

I don't dispute this at all. I have found, though, that a certain mix

[sdura...@aol.com]

In article <58pta8$c...@canyon.sr.hp.com>, Scott Frankland
<audi...@ix.netcom.com> writes:

>> ...transparent a window to the original event, would it not be
preferable
>> to put up with some bass distortion to get a better sense of
>> transparency?
>[snip -- bt]
>
>Well, why not have it all?

I wish that were possible. To my ears, though, as you reduce
distortion and flatten frequency response (as with sold-state
amplifiers), the sound becomes flatter, less three-dimensional and,
overall, less involving. Other SETA listeners report the same
results.

>> >They do--in SETA's. The SETA movement is to a large extent a movement
>> >away from accuracy as a criterion and toward beauty, emotion, and
>> >sheer involvement.
>>

>> I agree with this, except that "moving away from accuracy" is not
>> consistent with the greater sense of transparency that SETAs provide.
>> Transparency must be, by definition, more accurate.
>

>Not more accurate; more apparently real to a particular listener.

To a particular listener, wouldn't a sense of "realness" be consonant
with being accurate? Under what conditions could these two
perceptions diverge?

>> So what's at
>> issue is that SETAs are technically less accurate but perceptually
>> more transparent, which goes back to my whole sense that there's no
>> (or very poor) link between technical and perceptual accuracy,
>> published studies not withstanding.
>
>You're not exactly encouraging the process, here, Seigfreid.

Sorry, it's just my opinion and I know it's not very popular. No
offense, please.

>> ...there are too many examples,
>> both anecdotal and tested, of situations where distortion and
>> compression lead to a better sense of capturing the original event.
>
>Why not just make better recordings, microphones and speakers? Then
>we could all move toward pure transparency, i.e., an absense of all
>distortion products.

Again, refer to my comment above wherein SETA listeners have
(apparently) come to dislike the sound of low distortion. How could
this be?

[snip - bt]

[sdura...@aol.com]

In article <58ptel$c...@canyon.sr.hp.com>, dpl...@iq.navio.com writes:

>In this case, whose listening experience is valid? Whose preferred
>mix of linearities, nonlinearities, and distortions is correct?
>
>To my mind, the only answer is "Both are!" To say otherwise is to
>force people into a Procrustian bed, to no purpose other than the
>personal goal of being "right". Bah.

What you say makes a lot of sense, yet I can't help thinking that

there's some sort of universal truth to all of this. It's akin to
sounds that comfort humans regardless of culture. You play Mozart and
there tends to be a common response by people all over the world. Why
would this also not hold true for a sense of proper music
reproduction?

>>I agree with this, except that "moving away from accuracy" is not

>>consistent with the greater sense of transparency that SETAs provide.
>>Transparency must be, by definition, more accurate.
>
>I disagree. You are comparing apples and oranges here.

I've asked Scott the same question: To a given listener, why are

transparency and accuracy not consonant? Under what conditions would
these two perceptions diverge? What kind of perception is
"inaccurate" transparency? Wouldn't it no longer be transparent? Is
there a condition of accuracy that does not also convey transparency?

>I suggest the following: an increased sense of transparency and

>immediacy _is_ euphonic. You like it, you consider it to be an
>ultimate goal... clearly, it's euphonic (literally, "good sounding" I
>believe).

In an absolute sense, transparency and euphonics do not necessarily go

hand-in-hand. Clearly one can have transparent sound of something
that is not good-sounding (i.e. a recording of something that sounds
ugly) and it is no longer euphonic. Further, I have found that there
exists certain audio gear that makes things sound too good - overly
warm, recessed highs, blurred micro-detail, etc - that convey a sense
of euphony but is not transparent. Surely, you must have similar
experiences.

>I also suggest that some of the well-characterized euphonic

>distortions and nonlinearities (frequency-response irregularities,
>even-order harmonic distortion, dynamic effects, microphonics) may be
>responsible for at least part of what's contributing to the
>sense-of-transparency that you favor.

I don't dispute this at all. I have found, though, that a certain mix

[Steve Jones, contractor]

[Scott Frankland]

sdura...@aol.com wrote:
>
> >Well, why not have it all?
>
> I wish that were possible. To my ears, though, as you reduce
> distortion and flatten frequency response (as with sold-state
> amplifiers), the sound becomes flatter, less three-dimensional and,
> overall, less involving. Other SETA listeners report the same
> results.

Don't forget that the frequency response of a SETA is loudspeaker
dependent (i.e., it will be DIFFERENT for each loudspeaker). Besides,
f-response is not so much a spatial, as a tonal modifier. The
tonality of a given SETA is thus modified by the speaker to which it
is connected. What you are experiencing here is an EQUALIZED frequency
balance (obviously it is one that more closely matches your preference
than does a flat response).

The dimensionality illusion is more likely due to electrode resonances
than to even-order distortion products. Remember, these same
even-order products are present in solid-state SE amps, too. The main
effect of low order even-order products is to add body, richness, and
"beauty" to the sound. The intermediate products add
"transparency". Why do you go on denigrating the utility of technical
correlations? Until we can find correlations this whole process will
continue to be willy nilly.

> >Not more accurate; more apparently real to a particular listener.
>
> To a particular listener, wouldn't a sense of "realness" be consonant
> with being accurate? Under what conditions could these two
> perceptions diverge?

These two terms provide a means of distinguishing objective from
subjective. In that sense they have separate connotations. "Real"
means apparently real. Accurate means that it measures well. The fact
remains that we cannot measure a soundfield quite well enough (yet) to
satisfy human perception. But we go on trying. Why not encourage the
process?

> >Why not just make better recordings, microphones and speakers? Then
> >we could all move toward pure transparency, i.e., an absense of all
> >distortion products.
>
> Again, refer to my comment above wherein SETA listeners have
> (apparently) come to dislike the sound of low distortion. How could
> this be?

It be because the recordings are not perfect and neither are the
loudspeakers or the listening rooms. If they were, would you still
want distortion? Is there a "better than real"?

> From a consumer perspective, I perceive this as blindly
> groping.

You do realize that in order to provide such a test we will need to
find mo' better correlations between what is measured and what is
perceived? What you ask for in one breath you denigrate in the
next. Correlations are GOOD, Sig. Encourage them.

> Mmmm. You didn't answer the question. I don't mean to put you on the
> spot, Scott, but I would suspect that identifying an amplifier, blind,
> based on its performance plots is no easier than identifying the type
> of music on a CD by looking at it.

You are not even trying, here, Siegfried (and I am losing patience).
This is patently absurd. There are more correlations than you can yet
imagine. Not all are discovered yet, by any means, but to deny that
any exist, well, what can I say. Please see my upcoming Stereophile
article. The charts and graphs are included there. I can't reproduce
them in this environment or I would.

~SF~

[Joe Berry]

Scott Frankland <audi...@ix.netcom.com> wrote:

>If you look at those
>funny computer-enhanced stick people in the newer kiddy films with a
>subdued eye, they can almost come to life! OTOH, some of these
>photo-shopped pin-up girls (that look too-too perfect) are hard to
>criticize unless you really concentrate. Enhancement has its
>place. ;-)

My analogy to photography seems to have obscured the point I was
trying to make, which is simply that some "distortions" can be
perceived as legitimate enhancements--not because they hide blemishes
or add synthetic detail to the original, but because they make key
information in the original more apparent to the senses.

This may help to account for the preference for SE amplifiers over PP
even among some experieced listeners, who perhaps "should know
better."

JB

[Paul Guy]

With all this talk about SETA vs other forms of amps, I tried making
a single ended tube amp to compare it with the transistor amp that I
normally listen through (Linn LK100, modified).
My final version of a SE amp gives me 15 w/ch, has no electrolytics
in the individual final output supplies, uses modified PP transformers
to allow SE use, etc, etc. The unique property I designed into this
amp is its low output impedance without feedback. I can get about 0.5
ohms on the 6 ohm tap, or a 'damping factor' of about 10. With
feedback, the output impedance is about 0.1 ohms, the same as my solid
state amp.
The interesting point I'd like to make, is that in NO listening test
that I've conducted can the listener discern the difference between
the two amps when gains are matched. So far I've tried 8-10 listeners,
young and old, musicians, audiophiles. These tests are with feedback
ON. Many thought I was faking the switching, since the sound was SO
similiar. Most were quite surprised that with such divergent
technologies the performance was indistinguishable.
With feedback off, it's quite easy to tell the difference. That
difference appears to be a problem of frequency response due to the
speakers non-uniform impedance vs. frequency. In all tests without
feedback the listeners preferred the transistor amp. That was most
likely because the higher impedance of the tube amp output causes a
loss of mid-bass, and some peakiness at 8-10 khz. Those effects amount
to about 0.5 - 1 db, calculated from the impedance-freq. graphs of my
speakers (PSB Stratus Golds).
So a test to hear the effects of using a SETA amp could be
completely screwed up because of the impedance characteristics of your
speaker and the output impedance of the amp! I personally doubt if
there are any reasonable tests done to eliminate the effects of
speaker response. There are a few speakers designed for an almost
constant impedance vs. freq. and they would be a useful tool to test
with.
Personally, I believe, with all the many amps I've built, with all
the tests I've conducted, that ALL well designed amps will sound the
same, I don't care if they use tubes, transistors, magnetics,
pneumatics, switching amps etc. Once they are engineered properly to
reproduce the input signal at the output, you won't be able to tell
one from another. If they are designed deliberately or carelessly to
be flawed, then they will sound different, good or bad depends on your
own personal taste.
If you are interested in the design of my amps, I have circuits,
design and construction info that I can e-mail you back (they are way
to big to include here). After building something that is a throw-back
to the 1940's, my next project is to build a switching amp - something
a bit more up-to-date.

-Paul

[Scott Frankland]

Joe Berry wrote:

>
> Scott Frankland <audi...@ix.netcom.com> wrote:
>
> My analogy to photography seems to have obscured the point I was
> trying to make, which is simply that some "distortions" can be
> perceived as legitimate enhancements--not because they hide blemishes
> or add synthetic detail to the original, but because they make key
> information in the original more apparent to the senses.

Your analogy is well taken, and probably works just as you say it
does. Just remember, however, that when Western Electric set out to
make speech intelligible over long telephone lines they did not need
to resort to enhancement to make "key information more apparent to the
senses"; they needed only to eliminate the IMD and the crosstalk from
the repeater amplifiers!

They achieved this by REMOVING spurious information--by means of
negative feedback. Negative feedback was invented by WE mathematicians
during the '20s and '30s for the specific purpose of making human
speech more intelligible when electronically reproduced.

Now I ask you, did they succeed? Of course they did! Can speech be
made more intelligible yet? Mmmm, probably. But not by destroying the
basis of what has been built up and evolved over many decades; rather,
by building upon it. Speech intelligibility and tonality enhancement
are two different things, however. SE operation does less of the
former and more of the latter.

> This may help to account for the preference for SE amplifiers over

> even among some experieced listeners, who perhaps "should know

> PP better."

Like I said, pin-up girls DO benefit from those extra pixels here and
there. The question is, do you want real or do you want fantasy? SE
amplifiers will prettify recordings that don't need to be
prettified. They are not a universal panacea, in spite of Mr. Rankin's
siren calls to the contrary. They are an interesting alternative, and
should, IMO, be presented as such. Those with commercial investments
in SE may not agree with me, but they are simply wrong, as I have
shown.

~SF~

[Bob Trosper]

Sheldon D. Stokes wrote:

> The real evil element in ANY tube amp is those output transformers.
> The best transformers ever made were bad, and then they go down from
> there. Take a look at how bad a transformer butchers a signal, it's
> pretty scarey when you think about it.

> I still use tube amps and probably will until there are no more tubes,
> but I also know the limitations of the design type.

Hmnm. Perhaps that transformerless tube amp I've got at home is only a
figment of my imagination. Seriously, one can match the high-ish
output impedance of an OTL (output transformer-less) tube amp to the
high-ish impedance of some electrostatics and have a reasonably flat
response. And, of course, one doesn't have the flexibility of speaker
choice as one does with other less load-sensitive amps.

-- Bob Trosper

[sdura...@aol.com]

In article <591o68$r...@agate.berkeley.edu>, joeb...@atl.mindspring.com
(Joe Berry) writes:

>My analogy to photography seems to have obscured the point I was
>trying to make, which is simply that some "distortions" can be
>perceived as legitimate enhancements--not because they hide blemishes
>or add synthetic detail to the original, but because they make key
>information in the original more apparent to the senses.
>

>This may help to account for the preference for SE amplifiers over PP

>even among some experieced listeners, who perhaps "should know

>better."

It's a well-known effect to "burn-in" or slightly darken the periphery
of photographs to bring attention to the central part of the image
(where the subject is). Properly done, this causes the viewer to more
quickly understand the nature and meaning of the image. Is this not
the same effect as the SE amplifier?

An image that is not properly burned-in peripherally may be more
accurate to the negative, but it is less-successful at conveying the
image to the viewer. As an experienced viewer of photographs, images
that are not burned-in are often confusing as I'm engaged in
subliminally searching for the subject. Such is often the effect I
get from the sound of push-pull amplifiers - I'm often confused about
what the central focus of the musical message is.

Siegfried

[Steve Jones, contractor]

Scott Frankland wrote:
>
> Joe Berry wrote:
> >
> > Scott Frankland <audi...@ix.netcom.com> wrote:
> >

> > My analogy to photography seems to have obscured the point I was
> > trying to make, which is simply that some "distortions" can be
> > perceived as legitimate enhancements--not because they hide blemishes
> > or add synthetic detail to the original, but because they make key
> > information in the original more apparent to the senses.
>

> Your analogy is well taken, and probably works just as you say it
> does. Just remember, however, that when Western Electric set out to
> make speech intelligible over long telephone lines they did not need
> to resort to enhancement to make "key information more apparent to the
> senses"; they needed only to eliminate the IMD and the crosstalk from
> the repeater amplifiers!
>
> They achieved this by REMOVING spurious information--by means of
> negative feedback. Negative feedback was invented by WE mathematicians
> during the '20s and '30s for the specific purpose of making human
> speech more intelligible when electronically reproduced.

Just gotta add my 2 cents here.

IF the results today, listening in mono, in one ear, over a close
coupled transducer is the norm, I whole heartedly agree with adoption
of this groundbreaking (literally) research by WE (preceding *stereo*
by a decade). But the practical limitations of stereo were known and
addressed as a matter of technique (3 channel) as early as the
1950's. Too bad it was flushed in the translation to 2 channel home LP
reproduction. :-(

-Steve Jones

[sdura...@aol.com]

In article <591csi$j...@tolstoy.lerc.nasa.gov>, Scott Frankland
<audi...@ix.netcom.com> writes:

>The dimensionality illusion is more likely due to electrode resonances
>than to even-order distortion products. Remember, these same
>even-order products are present in solid-state SE amps, too. The main
>effect of low order even-order products is to add body, richness, and
>"beauty" to the sound. The intermediate products add
>"transparency". Why do you go on denigrating the utility of technical
>correlations? Until we can find correlations this whole process will
>continue to be willy nilly.

I'm sorry, I missed something here. What are the "intermediate"
products?
Are they the ones related to electrode resonances? Am I understanding
you correctly in that low-level, even-order distortion products add body
and richness and electrode resonances add transparency?

>> >Not more accurate; more apparently real to a particular listener.
>>
>> To a particular listener, wouldn't a sense of "realness" be consonant
>> with being accurate? Under what conditions could these two
>> perceptions diverge?

>These two terms provide a means of distinguishing objective from
>subjective. In that sense they have separate connotations. "Real"
>means apparently real. Accurate means that it measures well. The fact
>remains that we cannot measure a soundfield quite well enough (yet) to
>satisfy human perception. But we go on trying. Why not encourage the
>process?

I'm sorry. I'm not trying to "discourage" the process - I'm trying to
flesh it out better. This cannot always be done in an elegant,
gracious fashion. Please try to bear with me.

I understand that accuracy has a specific technical definition, but
I'm talking about the distinction between a sense of transparency and
a sense of accuracy from a perceptual basis. Looking at it from this
point of view, wouldn't you agree that one's sense of accuracy and
transparency are the same?

If a person's preference is reproduction that has the greatest sense
of transparency or "realness," they would also be sensing that the
reproduced sound is a very accurate facsimile of the original event.
I would very much encourage all listeners audition and discover gear
that best suits their sense of what is real and accurate.

>> Again, refer to my comment above wherein SETA listeners have
>> (apparently) come to dislike the sound of low distortion. How could
>> this be?
>
>It be because the recordings are not perfect and neither are the
>loudspeakers or the listening rooms. If they were, would you still
>want distortion? Is there a "better than real"?

I think if a person were to continually relate reproduced sound to
what they hear in the concert hall, they couldn't knowingly accept a
sound that is better than real. What kind of sound is a better
violin? A super-violin? Louder? More resonant? Less resonant?
Sweeter? In each case, the sound deviates from the aural memory of a
violin and must, by definition, be wrong.

This DOES NOT speak to the effect of flaws in aural memory or the
possibility of self-deception. But even under such conditions, we're
still talking about a person's perception of what they hear live.

In the last Stereophile, Gordon Holt talks of his hearing loss in the
top octave, but do you think he can no longer relate reproduced sound
to what HE hears in the concert hall? Of course he can.

>> Mmmm. You didn't answer the question. I don't mean to put you on the
>> spot, Scott, but I would suspect that identifying an amplifier, blind,
>> based on its performance plots is no easier than identifying the type
>> of music on a CD by looking at it.
>
>You are not even trying, here, Siegfried (and I am losing patience).
>This is patently absurd. There are more correlations than you can yet
>imagine. Not all are discovered yet, by any means, but to deny that
>any exist, well, what can I say. Please see my upcoming Stereophile
>article. The charts and graphs are included there. I can't reproduce
>them in this environment or I would.

First off, please, it's not that I'm not trying. Secondly, yes, I
agree there are SOME correlations, but there aren't enough to
adequately illustrate the different sounds of amplifiers. If there
were, there'd be NO market for subjective reviews and you could get
all the information you'd need from Popular Electronics.

Notice that my original query was very specific. I fully expect that
given a plot of a solid-state and tube amplifier, you'd correctly
guess which is which. But the efficacy of measured results begins to
break down pretty quickly after you get beyond a certain threshold.

Siegfried

[Scott Frankland]

sdura...@aol.com wrote:
>
> I'm sorry, I missed something here. What are the "intermediate"
> products?
> Are they the ones related to electrode resonances? Am I understanding
> you correctly in that low-level, even-order distortion products add body
> and richness and electrode resonances add transparency?

There are low-order products, high-order products, and intermediate
products. I doubt whether resonances "add" transparency; they probably
do enhance the sense of "air" between instruments, hall sound, and
other soundstage and imaging effects. Low-order products (especially
2nd order) tend to reinforce the fundamental, even filling in for it
in some cases (see Ladner, referred to in my original
post). Intermediate products (3rd, 4th, 5th) of bass fundamentals add
energy to midrange instrumental timbres, enhancing their
"presence". If the energy becomes too great, however, the timbre
becomes strident; and the effect is lost.

> I understand that accuracy has a specific technical definition, but
> I'm talking about the distinction between a sense of transparency and
> a sense of accuracy from a perceptual basis. Looking at it from this
> point of view, wouldn't you agree that one's sense of accuracy and
> transparency are the same?

I see it as analogous to one of those optical illusions, that, when
once you know the trick, it becomes easier to defocus and "find" the
illusion. I am trying to find true transparency. I am not interested
quite yet in playing mind games with myself. But this is a designer's
perspective. I don't blame you for seeking the ultimate illusion--
that's what it is anyway.

> I think if a person were to continually relate reproduced sound to
> what they hear in the concert hall, they couldn't knowingly accept a
> sound that is better than real. What kind of sound is a better
> violin? A super-violin? Louder? More resonant? Less resonant?
> Sweeter? In each case, the sound deviates from the aural memory of a
> violin and must, by definition, be wrong.

Careful. Every violin sounds different. And different again in each
hall. The memory is less impressed than the sensibility.

> Notice that my original query was very specific. I fully expect that
> given a plot of a solid-state and tube amplifier, you'd correctly
> guess which is which. But the efficacy of measured results begins to
> break down pretty quickly after you get beyond a certain threshold.

Exactly. When two amps sound indistinguishable to an experienced
listener, chances are they will measure very similarly (even if one is
tube and the other transistor; or if one is PP and the other is SE).
But this is true transparency--i.e., when they become INTERCHANGEABLE,
sound-wise and bench-wise. What you are talking about is enhancement,
not true transparency. Perceived transparency is a personal issue, and
so the term "accurate" used in that context is also personal. I prefer
to keep the terms as Webster has them.

~SF~

[Joe Berry]

sdura...@aol.com wrote:

>It's a well-known effect to "burn-in" or slightly darken the periphery
>of photographs to bring attention to the central part of the image
>(where the subject is). Properly done, this causes the viewer to more
>quickly understand the nature and meaning of the image. Is this not
>the same effect as the SE amplifier?

Sounds interesting, but I was actually thinking of a technique called
local contrast enhancement, which is used to sharpen the distinction
between adjacent objects in an image. This helps the eye make out
spatial relationships as well as resolve fine image detail.

I suspect that, by adding simple, in-phase harmonic distortion to a
musical waveform, SE amps do something analogous by disproportionately
amplify the parts of the waveform that inform the ear about the size
and location of acoustic events. Perhaps, for some listeners, this
helps make up for the imperfect nature of the recording process in
general, and of two-channel stereo home playback in particular.

My intent here isn't so much to advocate SE amps as to come to terms
with their popularity among even some knowledgeable listeners, despite
obvious technical limitations compared to PP. Given the differering
priorities among buyers of high end audio gear, it seems likely that
both types will coexist, and moreover, that designers of each type
will borrow what strengths they can from the other camp.

JB

[sdura...@aol.com]

In article <593vhj$j...@tolstoy.lerc.nasa.gov>, Scott Frankland
<audi...@ix.netcom.com> writes:

>Now I ask you, did they succeed? Of course they did! Can speech be
>made more intelligible yet? Mmmm, probably. But not by destroying the
>basis of what has been built up and evolved over many decades; rather,
>by building upon it. Speech intelligibility and tonality enhancement
>are two different things, however. SE operation does less of the
>former and more of the latter.

Mmmm. I understand vocal lyrics much more clearly since I've listened
to SETAs. Again, I'm not saying the referenced findings are wrong;
it's just that my experience is different.

Siegfried

[Pete Goudreau]

> On 13 Dec 1996, sdura...@aol.com wrote,

>
> In article <58ptel$c...@canyon.sr.hp.com>, dpl...@iq.navio.com writes:
>
> >In this case, whose listening experience is valid? Whose preferred
> >mix of linearities, nonlinearities, and distortions is correct?
> >
> >To my mind, the only answer is "Both are!" To say otherwise is to
> >force people into a Procrustian bed, to no purpose other than the
> >personal goal of being "right". Bah.
>
> What you say makes a lot of sense, yet I can't help thinking that
> there's some sort of universal truth to all of this. It's akin to
> sounds that comfort humans regardless of culture. You play Mozart and
> there tends to be a common response by people all over the world. Why
> would this also not hold true for a sense of proper music
> reproduction?

Likely because the former instance expresses the individual's response
to music and the latter, their response to sound...quite a bit
different perceptual focus. Personally, I can enjoy music I like
while listening to it on the phone while on hold. I daresay that
those that enjoy music couldn't care less how they hear it, only that
they do. Pardon me for jumping in but I just felt that this bears
repeating until the point is well taken...

> <snippage>

>
> >I also suggest that some of the well-characterized euphonic
> >distortions and nonlinearities (frequency-response irregularities,
> >even-order harmonic distortion, dynamic effects, microphonics) may be
> >responsible for at least part of what's contributing to the
> >sense-of-transparency that you favor.
>
> I don't dispute this at all. I have found, though, that a certain mix
> of distortions and non-linearities best suit my sense of transparency.
> Other mixes don't do it. Now I'm not arguing that my mix is best for
> everyone - it's just that it seems there's a considerable number of
> people out here who like different mixes. Which still makes me wonder
> what all this obsession with linear-ueber-alles is all about.

Well, that *is* an advancement...don't think me rude here but this
statement is quite the turnaround from a year or so ago where those of
us not using SETAs were considered by you as tin-eared slide rule
types, thanks very much for cutting us some slack.

Your perception of obsession is nothing more than your observation of
market forces at work, after all, if low distortion didn't sell, no
one would be losing money making it, would they? Don't confuse
perception with reality, especially in music reproduction...

Cheers,
Pete

[Steve Jones, contractor]

Scott Frankland wrote:

> Exactly. When two amps sound indistinguishable to an experienced
> listener, chances are they will measure very similarly (even if one is
> tube and the other transistor; or if one is PP and the other is SE).
> But this is true transparency--i.e., when they become INTERCHANGEABLE,
> sound-wise and bench-wise. What you are talking about is enhancement,
> not true transparency. Perceived transparency is a personal issue, and
> so the term "accurate" used in that context is also personal. I prefer
> to keep the terms as Webster has them.

I prefer the lexicon of manufacturing. The distinctions are important.

*Precision* and *accuracy* are two separate but not exclusive
expressions of random distribution.

How constrained or uniform a pattern of events are distributed, is by
definition: precise.

How the average distribution of a pattern of events is related to an
external reference, ie a 'Standard', is by definition: accurate.

Note: a set may be 'precise' in that individual members may be uniform
(matched assemblies), but fail to acchieve accuracy (missing the
mark). Or a set may be accurate (on average), but fail to be of
sufficient precision (uniformity) to acchieve interchangeablity.

What tolerance describes the window of uniformity at the standard of
accuracy that insures true 'interchangeablity'? Or more importantly,
what is the standard of accuracy?

Isn't it in the listening?

-Steve Jones

[Scott Frankland]

Steve Jones, contractor wrote:
>
> What tolerance describes the window of uniformity at the standard of
> accuracy that insures true 'interchangeablity'? Or more importantly,
> what is the standard of accuracy?
>
> Isn't it in the listening?

I don't see any purpose in equating accuracy with perceived realism,
considering that perceived realism is often an illusory quality. Why
not leave accuracy as Webster intended it? --i.e., as something
objectively verifiable under controlled conditions. There are plenty
of terms to describe the subjective response, but perhaps fewer that
describe an objective standard. "Accuracy" and "precision" are better
left for objective measurements, IMO. In any case, the degree of
perceived realism will be a personal response. And that is the main
point.

As far as "true interchangeability", yes, that is tantamount to
"sonically identical", IMO, because I don't find that we can yet
measure any two amplifiers and be certain that they will sound
indistinguishable. I don't think we're quite that good yet.

~SF~

[Stewart Pinkerton]

sdura...@aol.com writes:

>It's a well-known effect to "burn-in" or slightly darken the periphery
>of photographs to bring attention to the central part of the image
>(where the subject is). Properly done, this causes the viewer to more
>quickly understand the nature and meaning of the image. Is this not
>the same effect as the SE amplifier?

To a limited extent, this does provide an analogy.

>An image that is not properly burned-in peripherally may be more
>accurate to the negative, but it is less-successful at conveying the
>image to the viewer. As an experienced viewer of photographs, images
>that are not burned-in are often confusing as I'm engaged in
>subliminally searching for the subject. Such is often the effect I
>get from the sound of push-pull amplifiers - I'm often confused about
>what the central focus of the musical message is.

Having been hung several times in the Edinburgh Salon (deservedly,
some might say! :-)), I am more than familiar with the concentrating
effect of vignetting (a better description of the specific 'burning
in' effect you describe). It is important to remember that this is an
entirely artificial effect produced by the printer in order to direct
the viewers attention to his preferred area of the image. It has
NOTHING to do with fidelity to the original scene but is a
manufactured impression.

--

Stewart Pinkerton | If you can't measure what you're making,
A S P Consulting | how do you know when you've got it made?
(44) 1509 880112 |

"I canna change the laws o' physics" - the other Scotty

[sdura...@aol.com]

In article <596jp1$c...@canyon.sr.hp.com>, Scott Frankland
<audi...@ix.netcom.com> writes:

>There are low-order products, high-order products, and intermediate
>products. I doubt whether resonances "add" transparency; they probably
>do enhance the sense of "air" between instruments, hall sound, and
>other soundstage and imaging effects. Low-order products (especially
>2nd order) tend to reinforce the fundamental, even filling in for it
>in some cases (see Ladner, referred to in my original
>post). Intermediate products (3rd, 4th, 5th) of bass fundamentals add
>energy to midrange instrumental timbres, enhancing their
>"presence". If the energy becomes too great, however, the timbre
>becomes strident; and the effect is lost.

Thank you for your descriptions; they're very helpful.

So, in effect, we have low-order (2nd harmonic), intermediate-order
(3rd through 6th) and high-order (more than 6th). Then we have the
character of even-order (2nd, 4th, 6th) being more pleasing to the ear
than odd-order (3rd, 5th, 7th). Jean Hiraga suggests that the
distortion spectrum should decline from a 2nd harmonic reference at
the rate of 18dB/octave. Has this been your experience as well?

>I see it as analogous to one of those optical illusions, that, when
>once you know the trick, it becomes easier to defocus and "find" the
>illusion. I am trying to find true transparency. I am not interested
>quite yet in playing mind games with myself. But this is a designer's
>perspective. I don't blame you for seeking the ultimate illusion--
>that's what it is anyway.

Some time ago, someone here on r.a.h-e pointed out to me that the
process of recording robs instruments of harmonic detail. He suggested
that the SETA sound somehow "restores" some of the missing harmonics,
though obviously not in a complimentary way. Do you agree that detail
is lost in the recording process? Might not a reproduction scheme
that "adds" back some harmonic detail be preferable to one that adds
nothing? I would say that we can only justify seeking truly
transparent playback when we can accomplish a truly transparent
"capturing" of the original event.

>> I think if a person were to continually relate reproduced sound to
>> what they hear in the concert hall, they couldn't knowingly accept a
>> sound that is better than real. What kind of sound is a better
>> violin? A super-violin? Louder? More resonant? Less resonant?
>> Sweeter? In each case, the sound deviates from the aural memory of a
>> violin and must, by definition, be wrong.
>
>Careful. Every violin sounds different. And different again in each
>hall. The memory is less impressed than the sensibility.

Which sensibility? The one that appreciates the beautiful tone of the
violin, the one that appreciates how accurate the violin sounds or the
one appreciates the violinist's virtuosity? While I understand your
emphasis of sensibility (I am very much moved by the sense of beauty
conveyed by SETA amps), I am still unclear how it's any more reliable
than memory in evaluating the performance of a sound system.

Siegfried

[dpl...@iq.navio.com]

In article <594i61$m...@canyon.sr.hp.com> sduraybito writes:

>I'm sorry. I'm not trying to "discourage" the process - I'm trying to
>flesh it out better. This cannot always be done in an elegant,
>gracious fashion. Please try to bear with me.

>I understand that accuracy has a specific technical definition, but

>I'm talking about the distinction between a sense of transparency and
>a sense of accuracy from a perceptual basis. Looking at it from this
>point of view, wouldn't you agree that one's sense of accuracy and
>transparency are the same?

>If a person's preference is reproduction that has the greatest sense

>of transparency or "realness," they would also be sensing that the
>reproduced sound is a very accurate facsimile of the original event.
>I would very much encourage all listeners audition and discover gear
>that best suits their sense of what is real and accurate.

Because (as you've acknowledged) the word "accuracy" has a specific
technical definition, I think that you're making a mistake trying to
use the same word in a different sense (the "sense of accuracy from a
perceptual basis"). By overloading this term with two different (and
incompatible) meanings, I believe you're muddying the waters. In
effect, you're running the risk of following the example of Humpty
Dumpty - "When I use a word, it means just what I want it to mean".

I suggest the use of a different adjective - one which conveys the
meaning as you wish, without "colliding" with the well-established
technical term.

In thinking over what you've written, the word that I keep coming back
to is "convincing". I think it's a valuable adjective for a couple of
other reasons:

- It really seems to convey the sense of what you've been saying, and

- It acknowledges (as the word "accuracy" by itself does not) that
you're talking about a _subjective_ phenomenon which involves
yourself as an individual - your ears, brain, training, experience,
and preferences. It puts you in the equation.

>I think if a person were to continually relate reproduced sound to
>what they hear in the concert hall, they couldn't knowingly accept a
>sound that is better than real. What kind of sound is a better
>violin? A super-violin? Louder? More resonant? Less resonant?
>Sweeter? In each case, the sound deviates from the aural memory of a
>violin and must, by definition, be wrong.

You speak as if "the aural memory of a violin" were just one thing.
Bosh. Violins vary. The sound of an individual violin will vary
depending upon how it is played, the recording venue, etc.

As others have pointed out, you can't tell whether a particular
recording is accurate UNLESS YOU WERE THERE WHEN IT WAS RECORDED.
Even if you've heard numerous other performances of the same piece,
you won't know how the particular recording REALLY sounded unless you
were there.

Therefore: you cannot tell whether the recording (or the sound of it
as played through your system) is accurate. If you say "This is an
accurate recording" or "This is an accurate reproduction" based ONLY
on hearing the recording through your system, you're talking through
your hat. That's just the plain truth - you don't have enough facts
to make this judgement.

You can, however, judge how CONVINCING it is - because this is a
subjective experience. You can say "Yes, this sounds just like I
remember a good violin tends to sound. It convinces me that it's a
violin. Good show."

> >To my mind, the only answer is "Both are!" To say otherwise is to
> >force people into a Procrustian bed, to no purpose other than the
> >personal goal of being "right". Bah.

> What you say makes a lot of sense, yet I can't help thinking that
> there's some sort of universal truth to all of this. It's akin to
> sounds that comfort humans regardless of culture. You play Mozart and
> there tends to be a common response by people all over the world. Why
> would this also not hold true for a sense of proper music
> reproduction?

There's an overall tendency or trend, I agree. People tend to be
influenced by similar things, to at least some degree. On the other
hand, there is a great deal of variation, based on environment and
experience at the very least. For example, much Oriental music sounds
very strange to Western ears at first. A traditional piece of koto
music which could evoke strong emotion in a large percentage of
Japanese listeners, would probably leave most Americans quite cold.

I could mirror your argument in another domain: food. I could argue
that "There must be some sort of universal truth about what tastes
good. If you give people all over the world a taste of honey, they
tend to react in the same way. Why wouldn't it be universally true
for the tastes of all foods?". However, it's easily shown that people
have different (genetically-influenced) abilities to taste or
not-taste certain flavor substances, as well as different
preferences... and, thus, that foods some people like will be
distasteful to other people.

It's all well and good to look for a "universal truth". It's a nice
theory. It's misleading to assume that there MUST BE ONE, and that
it's an ABSOLUTE, and that it's SIMPLE... and it's arrogant to assume
that one's own experiences and preferences are necessarily a high
instantiation of that truth.

Theories MUST give ways to facts - otherwise, they're religious
declarations rather than theories. We've seen here some very factual
statements that certain experienced listeners find objectionable the
very same sonic characteristics that you find benign or even pleasant.
Those are facts. Whatever "universal truth" you seek to define MUST,
of necessity, encompass those facts - or it's bullpoop.

Where humans are involved, the only semi-universal truth I've run
across adds up to "It depends". Well, "Sh*t happens" comes close to
being universal too, I guess.

Yeah, it can be a cold, uncomfortable world out there when you don't
have some Absolute Truths to shield yourself with. Too bad. Get used
to it.

--
Dave Platt Speaker-to-kernels
dpl...@navio.com Navio Communications, Inc.
Visit the Jade Warrior home page: http://iq.navio.com/jade-warrior/

[Steve Jones, contractor]

Scott Frankland wrote:

> I don't see any purpose in equating accuracy with perceived realism,
> considering that perceived realism is often an illusory quality. Why
> not leave accuracy as Webster intended it? --i.e., as something
> objectively verifiable under controlled conditions. There are plenty
> of terms to describe the subjective response, but perhaps fewer that
> describe an objective standard. "Accuracy" and "precision" are better
> left for objective measurements, IMO. In any case, the degree of
> perceived realism will be a personal response. And that is the main
> point.

Last post on this thread (honest!)

How do we bend the linear scale of accuracy, in order to accomodate
the non-linear perception of human hearing (especially if stereo is
the medium)?

Accuracy implys nothing more than adherance to an external frame of
reference. Per Webster's: "'accuracy' 2b: degree of conformity of
measure to a standard or a true value-compare PRECISION
(itallics:Webster's) 2a." Is a Ford Taurus (a mass produced product)
more 'accurate' than a hand built AC Cobra? Yes, if your frame of
reference is in individual adherance to part metrology. No, if your
frame of reference is balls out FUN. Herein lies the paradox. How do
we corollate the frame of reference (accuracy) to what is *important*
to what we perceive?

Is the stereo paradigm valid if left uncompensated?

Note: these questions are rhetorical. NO follow up is implied, nor
warranted. Bye, bye!

-Steve Jones

PS- Great exchange Scott, and thanks for your patience. Posting on
these NGs IS the intellectual equivalent of being tied to a whipping
post. I certainly got a lot out of our discussion. Hope those rope
burns heal soon:-).

PPS- Cheers, Merry Christmas, Happy Hannukah, Unbridled Unmentioned
Holidays, and wishing ALL a Great New Year.

[brad sanders]

"Steve Jones, contractor" <sjo...@vcd.hp.com> sez:

> In the RDH4, they mention 170khz heating as a way to move this
> *possible* problem effectively out of band (audio). Just curious to try
> it, thats all.

I've heard one can achieve instant and powerful sexual gratification
by applying a cattle prod to an area around the male genitalia, too
-- but I'm in no hurry to try it...

FYI, RF (yes, 170KHz IS RF) in the heaters is generally a *bad* thing.
It raises the noise floor, increases internod, racks tuner
sensitivity, etc. I cannot imagine any of these effects NOT
propagating to "audio."

> In a 300B the filament is the cathode, and the inductance of the
> output transformer *may* help dampen the RFI from the PS. Don't know
> until I try it though.

Mebbe. Plate inductance does nothing to help this in RF circuits. You
can add as many pi stages to the output as you want, but you'll never
get rid of the noise and intemod you've added.

This is fun! Man, I love talkin' radio...

Disclaimer: The above does not necessarily reflect the
opinion of my unemployment counselor.
> http://radioactive.home.ml.org <------------------------------------

[Steven Abrams]

sdura...@aol.com writes:

> What you say makes a lot of sense, yet I can't help thinking that
> there's some sort of universal truth to all of this. It's akin to
> sounds that comfort humans regardless of culture. You play Mozart
> and there tends to be a common response by people all over the
> world.

Is this really true? Have there been any studies to show this?

I'm surprised, because Eastern music, played for Westerners for the
first time, (in my experience), generates a completely different
response than it does for Easternrers.

(Or are you somehow suggesting that there is a universality to Western
music ONLY?)

> I've asked Scott the same question: To a given listener, why are
> transparency and accuracy not consonant?

I think Gabe could (if he were still hanging out) answer this. Even
listeners who *prefer* systems with what we've often been calling
euphonic distortion agree that a good digital recording played via
quality solid-state electronics sounds closer to the direct studio
feed than an LP or a SETA system. It's a more *accurate* rendition.

Now, you get those people home with an LP and a CD, and I wonder if
some of them might still find that they can feel the music better via
the LP. They may feel that it renders a more transparent presentation.
They might also now feel that it sounds more realistic (and, to them,
more accurate). I bet that they would, since these are people who
prefer this presentation. No big deal if that's the case, that is just
their preference.

> I don't dispute this at all. I have found, though, that a certain mix
> of distortions and non-linearities best suit my sense of transparency.
> Other mixes don't do it. Now I'm not arguing that my mix is best for
> everyone - it's just that it seems there's a considerable number of
> people out here who like different mixes. Which still makes me wonder
> what all this obsession with linear-ueber-alles is all about.

Nobody has a linear-uber-alles obsession. Just keep in mind that
there are people who's prefered mix is NO distortions and NO
nonlinearities. And remember that they may well have come to this
conclusions using exactly the same methodology as you: listening to
lots of systems, not just looking up specs.

~~~Steve
--
Steven Abrams abr...@cs.columbia.edu

Living is easy with eyes closed, misunderstanding all you see.
-Lennon/McCartney

[jj, curmudgeon and all-around grouch]

In article <58ppb0$g...@agate.berkeley.edu>
stok...@sun.soe.clarkson.edu (Sheldon D. Stokes) writes:

> I'd look at the output transformer. You are fighting the laws of
> physics with SE output transformers. A good transformer should have
> very strong magnetic coupling between the primary and secondary
> (ref. radiotron). But with very good primary magnetic coupling to
> the core (and thus to the secondary), you run into core saturation
> problems when running DC through the primary as you have to do with
> all single ended amps.

So, why not just run a second winding on the transformer, and put a
current source into it that bucks the DC component of the output
stage?

I'll bet you can even find a transformer or three to use to do that.

<JJ cackles>

--
Copyright alice!jj 1996, all rights reserved, except transmission by USENET
and like facilities granted. This notice must be included. Any use by a
provider charging in any way for the IP represented in and by this article
and any inclusion in print or other media are specifically prohibited.

[brad sanders]

Bob Trosper <rtro...@hpsrjtc.sr.hp.com> sez:

>Hmnm. Perhaps that transformerless tube amp I've got at home is only a
>figment of my imagination. Seriously, one can match the high-ish
>output impedance of an OTL (output transformer-less) tube amp to the

>high-ish impedance of some electrostatics...

Yeah Bob - but then you gotta OTL amp - and a transformer coupled
speaker!

I don't think they're that much of a problem anyway; I just got a
chuckle out of the notion of someone going to the trouble to make an
OTL amp - and then couple it to those big ol' pieces of iron in the
back of a pair of panels...

---------------------------------- Inspired by actual events ----
> http://radioactive.home.ml.org <-------------------------------

[hen...@nortel.ca]

jj, curmudgeon and all-around grouch (j...@research.att.com) wrote:

: So, why not just run a second winding on the transformer, and put a

: current source into it that bucks the DC component of the output
: stage?

Interesting concept. You know, it occurs to me, as long as you
are going to the trouble of building that current source, you could
provide a means to modulate the current out of phase with the signal
driving the main output tube. In this way, you could augment the
total power delivered to the transformer primary instead of just
creating heat. You'd need some sort of 180 degree sign-swapper in
the signal path. Here's my idea:

input -> buffer -> sign-swapper (+) -> main output -> trans (+)
(-) -> current src -> trans (-)

I bet you could get twice the output power in Class A, and
even more if you drove the thing hard enough to shut off either
output tube during part of the signal cycle. I propose calling
this mode of operation "see-saw" mode because the signal goes
up on one end of the primary while it goes down on the other.

Has anyone ever considered this sort of circuit before?

-Henry

--
ATTENTION! Reply to h...@nortel.ca (hen...@nortel.ca won't work).

[sdura...@aol.com]

In article <599ih9$9...@agate.berkeley.edu>, dpl...@iq.navio.com writes:

>In thinking over what you've written, the word that I keep coming back
>to is "convincing". I think it's a valuable adjective for a couple of
>other reasons:
>
>- It really seems to convey the sense of what you've been saying, and
>
>- It acknowledges (as the word "accuracy" by itself does not) that
> you're talking about a _subjective_ phenomenon which involves
> yourself as an individual - your ears, brain, training, experience,
> and preferences. It puts you in the equation.

Yes, from a perceptual point-of-view, the term "convincing" can
accomplish a lot. I'm still trying to relate to the term
"transparency," because I think there are characteristics to this that
listeners can agree on. It's tougher to agree on "convincing,"
because, as you point out, it is so subjective. But I sense there's a
character to "transparency" that is more easily conveyed to another
person.

One of the sensations that means "transparency" to me is the aural
equivalent of being better able to see the soundfield and instruments
within. I can then point to something in the soundfield and relate
that "image" to another person.

When trying to convey "convincing" or "accurate," the visual sense is
somehow removed and so, to me, it's more difficult to share with
another person.

Just thinking out loud here.

>Therefore: you cannot tell whether the recording (or the sound of it
>as played through your system) is accurate. If you say "This is an
>accurate recording" or "This is an accurate reproduction" based ONLY
>on hearing the recording through your system, you're talking through
>your hat. That's just the plain truth - you don't have enough facts
>to make this judgement.
>
>You can, however, judge how CONVINCING it is - because this is a
>subjective experience. You can say "Yes, this sounds just like I
>remember a good violin tends to sound. It convinces me that it's a
>violin. Good show."

Yes, I get your point. I may even be "convinced" your right... ;-)

>Where humans are involved, the only semi-universal truth I've run
>across adds up to "It depends". Well, "Sh*t happens" comes close to
>being universal too, I guess.
>
>Yeah, it can be a cold, uncomfortable world out there when you don't
>have some Absolute Truths to shield yourself with. Too bad. Get used
>to it.

Well, I don't happen to agree. If we don't hold our perceptions up to
certain standards - the sense of a universal truth - then it's a
pretty chaotic free-for-all. Now I agree the universe is made up this
way, but that doesn't mean I have to accept that. Part of the
struggle for human existence is to go against the very tide of
entropy. Right or wrong, I believe there is an Absolute Sound that
systems should aspire to and I'll go to my grave trying to find it.
Else, why bother?

Siegfried

[Scott Frankland]

sdura...@aol.com wrote:

> Thank you for your descriptions; they're very helpful.

It helps to be encouraging, especially if you want straight answers!
<g>

> So, in effect, we have low-order (2nd harmonic), intermediate-order
> (3rd through 6th) and high-order (more than 6th). Then we have the
> character of even-order (2nd, 4th, 6th) being more pleasing to the
> ear than odd-order (3rd, 5th, 7th).

A quibble. Even-order products are not necessarily more consonant than
odd-order products. The key factor is that the COMBINATION of
fundamental with a particular harmonic is a CONSONANT interval.

> Jean Hiraga suggests that the distortion spectrum should decline
> from a 2nd harmonic reference at the rate of 18dB/octave. Has this
> been your experience as well?

You have not given a reference level. Thus, one could begin the
convergence at 20% 2nd harmonic, or at .02% 2nd harmonic. I submit
that this reference level will vary for each recording.

> Do you agree that detail is lost in the recording process? Might
> not a reproduction scheme that "adds" back some harmonic detail be
> preferable to one that adds nothing? I would say that we can only
> justify seeking truly transparent playback when we can accomplish a

> truly transparent "capturing" of the original event.

I believe I have already made such a statement in this thread. First
problem is, the "adding" of information is often heard as "noise" by
many listeners. Second problem is, the "information mix" will vary
from record to record. Digital processing may help here one day.

>> Careful. Every violin sounds different. And different again in each
>> hall. The memory is less impressed than the sensibility.

> Which sensibility? The one that appreciates the beautiful tone of
> the violin, the one that appreciates how accurate the violin sounds
> or the one appreciates the violinist's virtuosity? While I
> understand your emphasis of sensibility (I am very much moved by the
> sense of beauty conveyed by SETA amps), I am still unclear how it's
> any more reliable than memory in evaluating the performance of a
> sound system.

One's VALUE SYSTEM is what makes this determination, Siegfried. Your
emphasis is on beauty. So why not forget verisimilitude--go for
beauty! The sense of beauty, it seems to me, is an archetype; i.e., a
deep-seated, culturally engrained memory, inextricably combined with
one's native sense of hearing in this case. Thus arises one's
sensibility toward sound and music. One's memory of the concert hall
should not be relied upon as the sole paradigm for beauty because our
native sense of hearing UNDERLIES this experience. The concert hall
experience merely reinforces the archetype within. A different concert
hall, on planet Zenos, with alien instruments, might be just as
beautiful and moving as a night at Covent Garden! The symphony
orchestra is simply a manifestation of the WESTERN archetype.

OTOH, my emphasis is on verisimilitude. I want to try to recreate
exactly what the composer had in mind with HIS archetype. If resort
must be had to euphonic artifice in order to better divine his
purpose, so be it. I don't adhere to any "code of accuracy". I adhere
to my own sensibility. I use measurements and correlations to guide me
in this path. They are indispensable guideposts.

~SF~

[Scott Frankland]

sdura...@aol.com wrote:
>
> >Yeah, it can be a cold, uncomfortable world out there when you don't
> >have some Absolute Truths to shield yourself with. Too bad. Get used
> >to it.
>

> Well, I don't happen to agree. If we don't hold our perceptions up to
> certain standards - the sense of a universal truth - then it's a
> pretty chaotic free-for-all. Now I agree the universe is made up this
> way, but that doesn't mean I have to accept that. Part of the
> struggle for human existence is to go against the very tide of
> entropy. Right or wrong, I believe there is an Absolute Sound that
> systems should aspire to and I'll go to my grave trying to find it.
> Else, why bother?

The absolute sound is by definition an exact duplication of the
original sonic event. The problem is to produce it, and then to verify
it, both subjectively and objectively. With the advent of high speed
realtime analysis, we finally have at hand a means of corroborating
the subjective experience. It may take awhile, but as companies like
Audio Precision continue to evolve the state of the art in audio
metrology, we will continue to approach an independent, objective
means of verification.

I suspect that the ultimate analyzer will compare what is measured at
the opening of the ear canal in the concert hall vs. in an ideal sound
room. An ideal sound room, it seems to me, will be an anechoic
chamber. The speaker system will consist of perhaps as many as seven
loudspeakers, arrayed around the listening position. A computer will
adjust the signals in the time domain such that what arrives at the
ear canal will more closely resemble what was heard in the concert
hall.

An alternative system might, however, produce an even more compelling
experience. Sensors attached to the listener's skin will measure
endorphin levels, and the computer will adjust the sound according to
that. Interestingly, such a system may NOT necessarily reproduce the
absolute sound of the concert hall, because it is adjusting for
endorphin levels rather than for accuracy. Which, then, will you
prefer, the accurate system, or the one that sounds more compelling?

~SF~

[uh....OH!]

This is vearing off to metadiscussion again, but anyway ...

On Wed, 18 Dec 1996 10:32:54 -0800, "Steve Jones, contractor"
<sjo...@vcd.hp.com> wrote:

>Scott Frankland wrote:

>> I don't see any purpose in equating accuracy with perceived realism,
>> considering that perceived realism is often an illusory quality. Why
>> not leave accuracy as Webster intended it? --i.e., as something
>> objectively verifiable under controlled conditions. There are plenty
>> of terms to describe the subjective response, but perhaps fewer that
>> describe an objective standard. "Accuracy" and "precision" are better
>> left for objective measurements, IMO. In any case, the degree of
>> perceived realism will be a personal response. And that is the main
>> point.

>How do we bend the linear scale of accuracy, in order to accomodate

>the non-linear perception of human hearing (especially if stereo is
>the medium)?

I don't think we need to bend any scale to human perception,
particularly because if we aren't using accuracy as a reference, then
we're foregoing the measuring part, which is fine, and alleviates the
need for a scale.

> Accuracy implys nothing more than adherance to an external frame of
>reference. Per Webster's: "'accuracy' 2b: degree of conformity of
>measure to a standard or a true value-compare PRECISION
>(itallics:Webster's) 2a."

Accuracy implys adherance not to the frame of reference, but to
whatever objective truth is implied by the context of the use of the
term accuracy. We do compare performance to an external and
predictable frame of reference, or scale if you will, when we try to
evaluate how accurate something is. If we want to measure accuracy,
we make quantifiable comparisons to what we consider an accurate
representation. There comes the issue of choosing what to measure.
I'm not familiar with an AC Cobra, but

>Is a Ford Taurus (a mass produced product) more 'accurate' than a
>hand built AC Cobra? Yes, if your frame of reference is in individual

>adherance to part metrology. frame of reference is balls out

>FUN. Herein lies the paradox. How do we corollate the frame of
>reference (accuracy) to what is *important* to what we perceive?

A better analogy, or at least more accurate, is that a big fat
limosine will provide a smoother ride than a Ducati superbike, but
given a road of changing direction, such as highway 84 between 35 and
1, i.e. signal, at a constant speed of say 44.1 miles per hour, (for
the sake of simplicity and discussion, constant time, no gravity
wells, etc.) that Ducati will be able to more accurately track that
road, (stay on the right side). My analogy isn't perfect, because the
driver of the Ducati probably couldn't stay that slow for very long,
and the limosine would either slow down or hit a tree. Then again,
without a frame of reference, who's to say it isn't?

Exactly.

>Is the stereo paradigm valid if left uncompensated?

Do you wear sunglasses at night, so you can, so you can....

colin

[Stewart Pinkerton]

radio...@geocities.com (brad sanders) writes:

>Bob Trosper <rtro...@hpsrjtc.sr.hp.com> sez:

[ cut -- rgd ]

>Yeah Bob - but then you gotta OTL amp - and a transformer coupled
>speaker!
>
>I don't think they're that much of a problem anyway; I just got a
>chuckle out of the notion of someone going to the trouble to make an
>OTL amp - and then couple it to those big ol' pieces of iron in the
>back of a pair of panels...

While this is true, I have to admit (painful though it is for a
dyed-in-the-wool SS guy!) that the Tim de Paravicini designed
combination of tube amp directly driving original Quad ESL-57 speakers
does sound mightily impressive, leaving bass extension aside. No lumps
of iron in sight, just smooth clean music and the best I've heard from
these classic speakers. Yes OK it did have a mains transformer, but
you get the idea................

[sdura...@aol.com]

In article <59c8h3$6...@eyrie.graphics.cornell.edu>, Scott Frankland
<audi...@ix.netcom.com> writes:

>> Jean Hiraga suggests that the distortion spectrum should decline
>> from a 2nd harmonic reference at the rate of 18dB/octave. Has this
>> been your experience as well?
>
>You have not given a reference level. Thus, one could begin the
>convergence at 20% 2nd harmonic, or at .02% 2nd harmonic. I submit
>that this reference level will vary for each recording.

I would think that the 2nd harmonic should be as low as possible, but
below a certain point, it doesn't seem like it's possible to maintain
an 18 dB/octave spectrum. My understanding is that the problem with
amplifiers with vanishingly low THD levels is that odd-order
distortions are commensurately higher than 18 dB octave (i.e. higher
orders, especially odd ones, make up a significant portion of the
spectrum).

Purely from a layman's perspective, I would suspect that there's a 2nd
harmonic distortion level from where the 18 dB/octave spectrum can be
maintained, but that any higher level of 2nd harmonic would be
unnecessary. This is admittedly an engineering solution borne
entirely from "it looks elegant," but may very well be completely
undoable. Just food for thought...

>> Do you agree that detail is lost in the recording process? Might
>> not a reproduction scheme that "adds" back some harmonic detail be
>> preferable to one that adds nothing? I would say that we can only
>> justify seeking truly transparent playback when we can accomplish a
>> truly transparent "capturing" of the original event.
>
>I believe I have already made such a statement in this thread. First
>problem is, the "adding" of information is often heard as "noise" by
>many listeners. Second problem is, the "information mix" will vary
>from record to record. Digital processing may help here one day.

Until that day comes, to this listener, the SE triode amp compliments
the loss of harmonic detail in the recording process to a degree I've
never heard from other technologies. So, for now, it gets me closest
to a sense of what happened.

[snip - bt]

>The concert hall
>experience merely reinforces the archetype within. A different concert
>hall, on planet Zenos, with alien instruments, might be just as
>beautiful and moving as a night at Covent Garden! The symphony
>orchestra is simply a manifestation of the WESTERN archetype.

I had an interesting experience a couple of years ago. I came across
some recordings of Gamelan music that, I was assured, were of
reference quality. I spent some time listening to them, but just
couldn't cotton to the sound - too much midrange energy.

Then I has a chance to go to a live concert featuring Gamelan music
and heard things that enabled me to better understand how this music
works. Upon returning home, I played some of these Gamelan recordings
again and marveled at how much better they seem to sound.

>OTOH, my emphasis is on verisimilitude. I want to try to recreate
>exactly what the composer had in mind with HIS archetype. If resort
>must be had to euphonic artifice in order to better divine his
>purpose, so be it. I don't adhere to any "code of accuracy". I adhere
>to my own sensibility. I use measurements and correlations to guide me
>in this path. They are indispensable guideposts.

Just as you perceive our native hearing to underlie musical
sensibility (as you described above), I perceive the search for truth
to underlie the search for beauty. And as you correctly point out,
our musical sensibility can transcend our native hearing. So, too,
can the search for beauty transcend the search for the truth.

BTW, much of this "philosophisizing" that we've exchanged here has
been very thought-provoking for me. It has helped me better
understand not only the reasons for how people hear things, but also
how I hear things. In the spirit of the season, here's to you and
yours and I hope to continue our dialogue in the future.

Siegfried

[Scott Frankland]

sdura...@aol.com wrote:
>
> Yes, from a perceptual point-of-view, the term "convincing" can
> accomplish a lot. I'm still trying to relate to the term
> "transparency," because I think there are characteristics to this that
> listeners can agree on.

We can agree on it, we will just VALUE it differently depending
on our other priorities. You have made of it a touchstone. Others
might not, if something else they value is being traded-off in
the process.

> One of the sensations that means "transparency" to me is the aural
> equivalent of being better able to see the soundfield and instruments
> within. I can then point to something in the soundfield and relate
> that "image" to another person.

This particular effect lends excitement and interest to the experience
of listening to music, but it does not, IMO, contribute greatly to the
*beauty* of the sound. OTOH, it DOES contribute to ones involvement in
the performance! No doubt. Imaging does not suck.

~SF~

[Scott Frankland]

hen...@nortel.ca wrote:

> I bet you could get twice the output power in Class A, and even more
> if you drove the thing hard enough to shut off either output tube
> during part of the signal cycle. I propose calling this mode of
> operation "see-saw" mode because the signal goes up on one end of
> the primary while it goes down on the other.
>
> Has anyone ever considered this sort of circuit before?

I think Ed Colpitts used to call it "push-pull" way back in 1912.

~SF~

[Scott Frankland]

sdura...@aol.com wrote:

> I would think that the 2nd harmonic should be as low as possible, but
> below a certain point, it doesn't seem like it's possible to maintain
> an 18 dB/octave spectrum.

If it gets too low, you might miss the euphonics. ;-)

> My understanding is that the problem with
> amplifiers with vanishingly low THD levels is that odd-order
> distortions are commensurately higher than 18 dB octave (i.e. higher
> orders, especially odd ones, make up a significant portion of the
> spectrum).

I suspect that the way the ear overloads gives a clue as to what the
ideal descending order might be. I haven't calculated the roll-off
rates of the harmonic products but I DO see a remarkable resemblance
between the two spectra (i.e., the ear at clipping vs. SE at
clipping). I show comparative graphs in my Jan Stereophile article.

> Purely from a layman's perspective, I would suspect that there's a 2nd
> harmonic distortion level from where the 18 dB/octave spectrum can be
> maintained, but that any higher level of 2nd harmonic would be
> unnecessary.

Well, it can be maintained for ANY reference level. Problem is, at
some point the higher harmonics will get lost in the weeds!

> Until that day comes, to this listener, the SE triode amp compliments
> the loss of harmonic detail in the recording process to a degree I've
> never heard from other technologies. So, for now, it gets me closest
> to a sense of what happened.

Very well qualified, Siegfried. I can't argue with that (much as I
might want to).

> Then I has a chance to go to a live concert featuring Gamelan music
> and heard things that enabled me to better understand how this music
> works. Upon returning home, I played some of these Gamelan recordings
> again and marveled at how much better they seem to sound.

Precisely. And by the same token, the music on planet Zenos might
sound strange at first, but later....

> Just as you perceive our native hearing to underlie musical
> sensibility (as you described above), I perceive the search for truth
> to underlie the search for beauty.

OK, but this is a philosophical dilemma. My parallel is
psychophyisiological, and thus organic (innate). Because we have free
will, your parallel might go either way, depending on the individual.

> And as you correctly point out,
> our musical sensibility can transcend our native hearing. So, too,
> can the search for beauty transcend the search for the truth.

And sometimes the search for truth can lead to disaster, as it
did for King Oedipus.

> BTW, much of this "philosophisizing" that we've exchanged here has
> been very thought-provoking for me. It has helped me better
> understand not only the reasons for how people hear things, but also
> how I hear things. In the spirit of the season, here's to you and
> yours and I hope to continue our dialogue in the future.

Mirabilu dicta! And so greetings to one and all!

~SF~

[jj, curmudgeon and all-around grouch]

In article <599ina$a...@agate.berkeley.edu> "Steve Jones, contractor" <sjo...@vcd.hp.com> writes:
>How do we bend the linear scale of accuracy, in order to accomodate
>the non-linear perception of human hearing (especially if stereo is
>the medium)?

Why do we "bend" it due to the non-linearity of the ear? The same ear
listens to both the signals, so the same non-linearities are present
in both.

Now, the soundfield issues are something else, there is NO accuracy to
speak of in "soundfield restoration" in stereo, so it is reasonable,
at present, to play with something else, but what this is varies
between listeners, and with listener experience, so again, there's no
"universal".

[c...@bga.com]

In <599ub2$4...@eyrie.graphics.cornell.edu>, j...@research.att.com (jj, curmudgeon and all-around grouch) writes:
>In article <58ppb0$g...@agate.berkeley.edu>

>So, why not just run a second winding on the transformer, and put a
>current source into it that bucks the DC component of the output
>stage?

Isn't the DC component in the transformeer of a no-feedback SETA
actually an advantage in that it biases the operating point of the
transformer away from the flux zero crossing point? That way, it
avoids the relatively strong hysteresis at that point that would
introduce odd harmonic components that do not scale down with signal
strength.

Moderator Note: The DC flux is an advantage as you claim; it's a
dissadvantage in that it will saturate the core more easily. Life is
always a tradeoff. Hope I got that right Mike L. -- rgd ]

--
Cheers,

Chris. Broadbent ( KC5VQL )

[Scott Frankland]

c...@bga.com wrote:
>
> Isn't the DC component in the transformeer of a no-feedback SETA
> actually an advantage in that it biases the operating point of the
> transformer away from the flux zero crossing point? That way, it
> avoids the relatively strong hysteresis at that point that would
> introduce odd harmonic components that do not scale down with signal
> strength.

Well, you can always UNBALANCE a push-pull amp to get up over the
reluctance hump. You can also interleave core materials (such as 78%
nickel Permalloy) into the stack that energize at darn near .03
Oersteds. I was expecting this argument at the beginning of the thread
but none of the SE gurus brought it up. So I didn't either. Why should
I help them? They started the argument, not me.

~SF~

[Michael S. LaFevre]

Scott Frankland wrote:
>
> c...@bga.com wrote:

[ cut -- rgd ]

Scott Frankland wrote:
> Well, you can always UNBALANCE a push-pull amp to get up over the
> reluctance hump. You can also interleave core materials (such as 78%
> nickel Permalloy) into the stack that energize at darn near .03
> Oersteds.

Intentionally using unbal DC in a trans not designed for such use is
not IMO the path to optimality. The unbalanced DC creates a DC flux
which decreases the amount of flux capacity left for the AC chores.
In a trans designed to carry DC unbal plate current generally speaking
the cores will be physically much, much larger than an equivalent PP
trans would be for the same AC power rating and the SE trans will
generally employ an air gap to keep the DC flux within bounds (or so
we hope).

The "nickel" trick in PP trans...works...with the cavaet that above a
certain operating level it goes into saturation. Peerless utilized
this strategy in some of their output transformers...but again no free
lunch at 15 Kg flux density the nickel is let us say not a
particularly happy camper...obviously each designer weighs and makes
trade-offs in their designs...and far be it for me in particular to
criticize the elegant engineering solutions\approaches that the
original design engineers at Peerless settled on since much of my
learning of magnetics has been derived just exactly from this body of
work. But on balance it is not an approach that I really feel super
comfortable with.

An interesting paper and approach to air gaps and unbal plate currents
in tranneys was authored by the late Dr. Partridge in 1939. It
consisted of a four part series published in Wireless World. In the
last installment Dr. Partridge presents some detailed arguments for
the use of air gaps in PP transformers (which do not by design carry
any DC unbal plate currents).

Now I sure as hell hope that I am not a SE guru (or any other such
foolishness). I've steered clear of these related threads because they
have looked a bit like a bait trap to me...and personally I don't feel
like arguing audio till the cows come home. Like McDonald's sez folks
can "have their way".

Happy new year everyone,

Mike LaFevre

[Scott Frankland]

Michael S. LaFevre wrote:

> Intentionally using unbal DC in a trans not designed for such use is
> not IMO the path to optimality. The unbalanced DC creates a DC flux
> which decreases the amount of flux capacity left for the AC chores.

So who said a C-core was illegal for PP?

> But on balance it is not an approach that I really feel super
> comfortable with.

Neither do I, but I thought I'd mention it.

> An interesting paper and approach to air gaps and unbal plate currents
> in tranneys was authored by the late Dr. Partridge in 1939. It
> consisted of a four part series published in Wireless World. In the
> last installment Dr. Partridge presents some detailed arguments for
> the use of air gaps in PP transformers (which do not by design carry
> any DC unbal plate currents).

Air gaps can be beneficial, if used with discretion. They can help to
linearize the inductance vs. DC level.

> Now I sure as hell hope that I am not a SE guru (or any other such
> foolishness). I've steered clear of these related threads because they
> have looked a bit like a bait trap to me...and personally I don't feel
> like arguing audio till the cows come home. Like McDonald's sez folks
> can "have their way".

No one is baiting anybody on my end. I'm just pointing out the design
challenges in EACH technology. BTW, I have nowhere in this thread
expressed a preference for EITHER technology. I like 'em both (for
different reasons)!

> Happy new year everyone,

Ditto!

~SF~

[Michael S. LaFevre]

Scott Frankland wrote:
>
> Michael S. LaFevre wrote:
>
> > Intentionally using unbal DC in a trans not designed for such use is
> > not IMO the path to optimality. The unbalanced DC creates a DC flux
> > which decreases the amount of flux capacity left for the AC chores.

> So who said a C-core was illegal for PP?

No one said they are illegal...nor did I see you mention it before now
and I of course made no mention of it. But a C-core without an air gap
will still exhibit the saturation properties of the core materials
which it employs if it is "pushed" into service having to handle
unbalanced DC and it was not designed to do such in the first place.

> > But on balance it is not an approach that I really feel super
> > comfortable with.
>
> Neither do I, but I thought I'd mention it.

Maybe here is my chance to correct a mistatement that I made
inadvertently in my first post on this thread. I missed in my initial
reads that you were saying to use 78% nickel as an adjunct or
admixture in the core....and I said Peerless did this. They did but
not with high nickel ever but rather with the use of 50% nickel (low
nickel). The high nickel saturates way too easy (in these
stereotypical applications...namely large signal output transformers).

> > An interesting paper and approach to air gaps and unbal plate currents
> > in tranneys was authored by the late Dr. Partridge in 1939. It
> > consisted of a four part series published in Wireless World. In the
> > last installment Dr. Partridge presents some detailed arguments for
> > the use of air gaps in PP transformers (which do not by design carry
> > any DC unbal plate currents).
>
> Air gaps can be beneficial, if used with discretion. They can help to
> linearize the inductance vs. DC level.

But this misses the core (pardon the pun) of Partridge's thesis...he
argues for the use of air gaps in push-pull tranneys that carry *no*
unbal plate currents. In other words he recommends air gapping a trans
independently of whether or not there is *any* unbal DC energizing the
core or not. You should check out this article in it's entirety...it
is a landmark article.

> > Now I sure as hell hope that I am not a SE guru (or any other such
> > foolishness). I've steered clear of these related threads because they
> > have looked a bit like a bait trap to me...and personally I don't feel
> > like arguing audio till the cows come home. Like McDonald's sez folks
> > can "have their way".
>
> No one is baiting anybody on my end. I'm just pointing out the design
> challenges in EACH technology. BTW, I have nowhere in this thread
> expressed a preference for EITHER technology. I like 'em both (for
> different reasons)!
>

As well I just enjoy building tranneys...don't really care if they are
push-pull critters or SE guys. My own preference for "at home"
listening is SE...at least until I build a low power 845 PP amp :=).

Mike

[Scott Frankland]

Further to my previous post, I here propose to evaluate, in a very
general way, the sensitivity of class AB amplifiers to output
transformer leakage reactance. This sensitivity, once established,
will then be compared to that of the single-ended amplifier.

In a push-pull amplifier, class AB means that there is a certain
amount of overlap between the half-cycles about the zero-crossing. The
amount of idle current is adjusted so as to overcome crossover
distortion at low power levels. Unfortunately this is not usually
enough to prevent slewing-induced distortion (SID) at high power
levels (in tube-type amplifiers that incorporate output transformers).

SID affects the top end at the higher power levels whenever the load
contains a capacitive element. This is independent from
first-stage-related SID (usually associated with poorly designed
feedback amplifiers). In transformer-coupled amplifiers,
output-stage-related SID is proportional to the interwinding
capacitance of the transformer. Since the tubes must charge this
capacitance, the load on the tubes increases as the amount of charge
transferred per unit time increases.

The load is determined by the impedance across the circuit, in this
case capacitive reactance, Xc. At any given frequency Xc can be
determined from

Xc = 1/(2pi x fC).

This equation states that the impedance drops as frequency f increases
(for a fixed C) [1].

The work of transferring charge around the cap is defined as

W = Q^2/2C,

where W = total work done; Q = total charge transferred; and C =
capacitance [2]. Since Q is the product of I x T, where I = current
and T = time, the amount of work required per second to transfer a
given amount of charge increases with frequency because at higher
frequencies there are more charging cycles per second, therefore more
total charge is transferred per second.

It is known that charge is moved whenever the voltage across the cap
*changes*. The amount of charge transferred per unit time is defined
as capacitive current ic where

ic = C(dv/dt).

The changing voltage thus does the *work* of transferring charge
around the cap. This becomes apparent when we consider that the slope
of the potential line (voltage distribution) across the cap at any
given point is equivalent to the electric field intensity at that
point [3], and that the field intensity is what holds the charge in
place. The slope (rate-of-change) of the potential line is thus
proportional to the force on the charge at any given instant. The
faster the voltage waveform changes, therefore, the greater the field
intensity across the cap becomes.

The amount of work required also increases as the tube voltage swing
increases since

W = QV/2.

The increase in work with either a frequency or a voltage increase
explains the physical basis for slew-rate limiting in amplifiers. The
slew-rate Sr for any amplifier is given by

Sr = 2pi x fmax x Vp,

where Vp is the peak voltage. The result is then divided by 1,000,000
to obtain the slew-rate in volts/microS. The most honest way to
specify the slew-rate is to take the amp up to its maximum peak
voltage at midband, and then to sweep upward in frequency until the
sinewave just takes on a triangular shape. The slew-rate limit can
then be calculated at this maximum frequency, fmax.

Since the tube is in series with the transformer primary, any increase
in interwinding capacitive current ic will flow through the tube. When
this happens, the load-line seen by the tube tilts upward along the
current axis. Thus, we can say that, the faster the rate of voltage
change, *or* the greater the voltage *swing*, the more difficult the
load on the tube becomes.

The most important thing that can be done to reduce the work load at
high frequencies is to reduce the interwinding capacitance of the
transformer--as by interleaved or bifilar winding configurations. Most
high quality output transformers incorporate these techniques, whether
intended for push-pull or for single-ended amplifiers.

In push-pull amps, a further improvement can be obtained by increasing
the idle current, Ib , which then moves the class of operation toward
class A. Because plate current does not cut off in a class A
amplifier, the interwinding capacitance of a given half-primary is
never fully discharged; therefore less total charge is transferred
throughout any given cycle. This translates into less total work done
per cycle.

The worst case SID occurs in class B amplifiers because here the
interwinding capacitance must be ramped up from empty (since class B
operates the tubes at zero idle current). This is consistent with the
charge integral equivalent of work

W = the integral from 0 to Q of q dq .

Class AB amps will fall somewhere in between the two extremes of class
A and class B, depending on the idle current setting of the particular
amplifier. >From the tube's point of view there is also a better
impedance match in class A because the conditions for impedance
matching will be more nearly met at each instantaneous point along the
operating cycle [4].

It is clear, therefore, that in a transformer-coupled push-pull
amplifier the class of operation can have a profound effect on the
quality of the top end; especially for amplifiers with highly
capacitive output transformers or low idle currents.

A similar problem occurs when the transformer possesses excessive
leakage inductance. For class B amplifiers, or for those biased near
class B, notch distortion occurs at the zero-crossing. The distinction
between notch distortion and crossover distortion is that notch
distortion increases with power output, while crossover distortion
decreases with power output. There is thus a form of zero-cross
distortion that occurs at both extremes of the power band.

The notch reflects the effect on the waveform of high-order distortion
products. The notch occurs because, in a class B amplifier, the plate
current slams off (and on) at the zero-crossing. The sudden transfer
of current (what Sah called "quasi transients") from one primary half
to the other exites a resonance between the interwinding capacitance
and the leakage inductance [5,6].

The solution to notch distortion is exactly the same as that for
excessive interwinding capacitance: (1) reduce the leakage inductance
by skillful transformer design; and/or (2) increase the idle current
so as to enrich the class AB operation. Even with these constraints, a
push-pull amplifier will have less total distortion at the top-end
than will a single-ended amp, in spite of the fact that single-ended
amps necessarily operate in class A. That is because in most cases
there will be less leakage reactance in a push-pull transformer [7].

~SF~

REFERENCES

[1] Bernard Grob, *Basic Electronics*, 1965.

[2] Sears, Zemansky, and Young, *University Physics*, 1976.

[3] William G. Dow, *Fundamentals of Engineering Electronics*, 1937.

[4] H.L. Kraus, "Class-A Push-Pull Amplifier Theory," Proc. IRE, Jan
1948, pp. 50-52.

[5] A. Pen-Tung Sah, "Quasi Transients in Class B Audio-Frequency
Push-Pull Amplifiers," Proc. IRE, 24:11, Nov 1936, pp. 1522-1535.

[6] Norman H. Crowhurst, "Realistic Audio Engineering Philosophy",
Audio, Oct. 1959, pp. 52-60, cont. pp. 113-114.

[7] Staff of the Department of Electrical Engineering: M.I.T.,
*Magnetic Circuits and Transformers*, 1958.

[Scott Frankland]

Michael S. LaFevre wrote:
>
> But this misses the core (pardon the pun) of Partridge's thesis...he
> argues for the use of air gaps in push-pull tranneys that carry *no*
> unbal plate currents. In other words he recommends air gapping a trans
> independently of whether or not there is *any* unbal DC energizing the
> core or not. You should check out this article in it's entirety...it
> is a landmark article.

I've seen it (briefly), and am suitably impressed. What I am saying,
however, is that PP amps can take advantage of air gaps in the same
way that SE amps do. They can be designed as you say to accommodate DC
flux with room left over for AC peaks, just as in SE amps. NOW we can
unbalance the PP halves to overcome the characteristic reluctance hump
of the iron. A few lams of nickel of suitably chosen alloy can be
inserted to further narrow the gap. Now where's the problem?

> As well I just enjoy building tranneys...don't really care if they are
> push-pull critters or SE guys. My own preference for "at home"
> listening is SE...at least until I build a low power 845 PP amp :=).

Again, my aim is to continue to evolve both techologies in parallel,
with full recognition of the technical difficulties involved in each
case.

~SF~

[Scott Frankland]

Scott Frankland wrote:
>
> It is clear, therefore, that in a transformer-coupled push-pull
> amplifier the class of operation can have a profound effect on the
> quality of the top end; especially for amplifiers with highly
> capacitive output transformers or low idle currents.

I will now summarize the implications of my recent post on amplifier
sensitivity to output transformer leakage reactance. This post is
specifically directed at transformer designers. I think the bottom
line should be clear by now: minimize interwinding capacitance above
all. The leakage inductance is far less critical so long as enriched
class AB operation is anticipated (class A operation is not necessary
in push-pull amplifiers for reasons which I won't go into here).

Leakage inductance is relatively benign in such amplifiers considering
that the power sensitivity of tube amplifiers to an impedance match is
much less critical in the direction of a rising impedance (as for
inductive loads) relative to a falling impedance (as for capacitive
loads) [1]. Obviously, it would be most desirable to reduce both types
of reactance, but that would require eliminating the output
transformer (a whole 'nother can 'o worms).

~SF~

[1] A rare plot of the real-world power sensitivity of tube amplifiers
is given in *Principles of Electronics and Electronic Systems*, by
J.L. Daley, ed., United States Naval Institute, Annapolis, MD, 1956.

[Paul Guy]

Scott Franklin wrote a long treatise on why capacitive effects of the
output transformer were aggravated by running class AB or B. It was
quite an article, I got lead down the path for quite a way before I
started to cuss. Scott must write articles for scientific journals -
if he hasn't, he sure has the gift.

Anyhow, without too much quoting, here's the part that I found really
puzzling (after wading through all the formulae):

>In push-pull amps, a further improvement can be obtained by increasing
>the idle current, Ib , which then moves the class of operation toward
>class A. Because plate current does not cut off in a class A
>amplifier, the interwinding capacitance of a given half-primary is
>never fully discharged; therefore less total charge is transferred
>throughout any given cycle. This translates into less total work done
>per cycle.
>
>The worst case SID occurs in class B amplifiers because here the
>interwinding capacitance must be ramped up from empty (since class B
>operates the tubes at zero idle current). This is consistent with the
>charge integral equivalent of work

....[snip]....

>It is clear, therefore, that in a transformer-coupled push-pull
>amplifier the class of operation can have a profound effect on the
>quality of the top end; especially for amplifiers with highly
>capacitive output transformers or low idle currents.

Now here's the rub. If one side of the winding has zero current (as
with class AB or B circuits, WHAT ABOUT THE OTHER SIDE? Remember the
transformer windings in a PP transformer are COUPLED together. Also
remember this is an analog device not a switch. If you are considering
slew rate performance from the consideration that the tubes are
switches, draw up the equivalent circuits and analyse them. I think
you'll find that the bias has very little to do with switching speed
(slew rate). Please note, that I'm not talking about inductive
effects! In that case bias current will have an effect (as far as
switching characteristics are concerned) because of energy stored in
the core. I'm also ignoring the effects of transconductance changing
over the operating conditions.

Quite apart from all the formulae, my own experiences with SE and PP
trannies is that you want leakage inductance and capacitance reduced,
the compromise comes from the resonant effects, and plate and load
impedances at high frequencies. If you measure an UNLOADED plate
transformer impedance vs freq., the impdedance is inductive (and
impedance linearily increasing) until a few kiloherz, and then
capacitive (impedance linearily decreasing) with higher frequencies.
Past about 30-40KHz, there are all sorts of bumps and hiccups from
stray interactions. You can also do the same test with the secondary
shorted. Then the impedance is very low until 10-20 KHz, when it
starts to rise from leakage inductance. This is a bit irrelevant
but....

>A similar problem occurs when the transformer possesses excessive
>leakage inductance.

...[snip]....

>The notch reflects the effect on the waveform of high-order distortion
>products. The notch occurs because, in a class B amplifier, the plate
>current slams off (and on) at the zero-crossing. The sudden transfer
>of current (what Sah called "quasi transients") from one primary half
>to the other exites a resonance between the interwinding capacitance
>and the leakage inductance [5,6].

Probably more to the point, it's very difficult to run class B. A poor
implementation will run marginal class C, where a small chunk of the
waveform is lost. This is not an issue with class AB, and only of
interest to those very few that are running good class B (like you can
do with transistors and get very low distortion).

>The solution to notch distortion is exactly the same as that for
>excessive interwinding capacitance: (1) reduce the leakage inductance
>by skillful transformer design; and/or (2) increase the idle current
>so as to enrich the class AB operation.

The McIntosh amps are an example of (1). These were run class B, but
utilized transformers that had very low leakage inductance (I believe
they were wound trifilar or multi-filar, with the 2 plate windings,
the output winding, and feedback windings - global and screen)

>push-pull amplifier will have less total distortion at the top-end
>than will a single-ended amp, in spite of the fact that single-ended
>amps necessarily operate in class A. That is because in most cases
>there will be less leakage reactance in a push-pull transformer [7].

Naah, that's just too general, and you've left too many loopholes to
make a good argument. I have a SE tube amp running right here that has
very low distortion! Now it doesn't have the power capability and
distortion of the same tubes run at the same conditions in PP, but it
has almost nothing to do with the transformer leakage inductance.
Until you run a PP amp right into class B, or overdrive the SE stage,
the direct affects of winding capacity and inductance aren't a real
cause of distortion. Indirectly, the design of the transformer can
influence distortion (especially at low frequencies), especially when
the magnetizing inductance isn't large enough, or has linearity
problems from saturation effects.

Even the effects you claim are happening, how significant are they? I
have yet to see them. Certainly they are not in the top 20 concerns
when optimizing a tube amp. Maybe from a frequency response, or phase
margin point of view, but distortion? Do you have numbers? This should
be an easy experiment to verify, just add some parasitic capacitance
via discrete caps hung off the primary. Then measure high and low
level distortion. Please don't make this another one of those horrible
audio myths!

-Paul

[Michael S. LaFevre]

Scott Frankland wrote:

> I've seen it (briefly), and am suitably impressed. What I am saying,
> however, is that PP amps can take advantage of air gaps in the same
> way that SE amps do. They can be designed as you say to accommodate DC
> flux with room left over for AC peaks, just as in SE amps. NOW we can
> unbalance the PP halves to overcome the characteristic reluctance hump
> of the iron. A few lams of nickel of suitably chosen alloy can be
> inserted to further narrow the gap. Now where's the problem?

What throws me a bit is the notion that PP amps could take advantage
of air gaps in the same way as SE amps do...I am not exactly sure what
your getting at here. SE amps use an air gap for a very, very
practical reason...that is that they carry significant amounts of
unbal plate currents and if the trans did not have an air gap it would
be severly saturated. PP amps do not enjoy this problem (at least to
the extent that a SE circuit does)...and I hope no one would want to
impose it on a PP amp...it would imply that the halves of the circuit
would be operating very, very non-symmetrically...PP as I understand
it aims at making the conditions of operation as close as is
practically obtainable for the push and the pull sides of the circuit
to each look (electrically) and behave similar (or identically). Equal
but opposite is what I thought we aim for in PP circuit parameters and
design. You know circuits and their analysis a lot better than
me...isn't it seeking difficulties to deliberately unbalance the plate
currents of each half? Wouldn't this also mean that the grid voltages
would now be different for each half and each half would require
differing amounts of drive to swing the grids? If so...then this
really does not seem to be the optimal path to pursue in a PP
strategy. So independently of whether the trans could be designed to
handle unbal DC plate currents (we know that they can at a price) is
this an appropiate strategy?

Regarding the "reluctance hump" of electrical grade of lamination
material. As Partridge points out in his article inductance varies
quite a bit in PP trans depending on the amount of drive being applied
to the primary windings. There is a spot a "reverse knee" might we
say at the bottom of the BH curve but the use of an air gap by itself
doesn't get us over this knee per se...might actually make it more
difficult to overcome...by forestalling it by the demand for more
exciting force to be applied since with an air gap some of the
exciting force is expended in the air gap itself. Partridge
recognizes this explicitly. The chief advantage in air gapping a PP
is that you can make the inductance much flatter in magnitude vs the
exciting force applied to the iron and obtain more nearly the "same"
inductance over a much wider range of exciting forces. This differs
from SE in so far as SE circuits typically impose a large (relatively
speaking) DC unbalanced plate current through the iron...and yep this
does "bias" the iron upwards further into the BH curve than a PP
ciruit would do. And this may be of benefit if properly accounted for
in the design of the magnetic circuit...but it can also be an
albatross if not conservatively designed for in the iron
circuit...then the DC flux consumes the linear region of the BH curve
and the AC signal volts (the music) must settle for the non-linear
region of the BH curve up around the knee of the curve or beyond.
There are many SE tranney designs out there spec'd for massive amounts
of DC plate current that leaves no appreciable room for the AC flux
and saturates when you ask it to do low frequency signal reproduction
at any appreciable power level. No one should want to replicate this
behavior.

Regarding nickels...yes they have much, much higher initial perms than
the silicon steels. But the transformer gods declared that there
would be no free lunch. So the fly in the free soup is that the
nickel materials saturate very early vis-a-vis the "ordinary" grades
of steel like the M6 material. It's a trade at best...

I very much respect the work Partridge did and he was very thorough
for the most part. His strategy works but again with an expense of
increasing the exciting current which itself limits the ability of the
iron to deliver clean power at low frequency signals. I have a paper
by Ercel B. Harrison on this. If you would like a copy please let me
know.

Mike

[Paul Guy]

"Michael S. LaFevre" <m...@magnequest.com> wrote:

>
>What throws me a bit is the notion that PP amps could take advantage
>of air gaps in the same way as SE amps do...I am not exactly sure what
>your getting at here.

....[snip!]....

By having an air gap in a PP transformer (I've done this with a
hacksaw to a nice pair of Partridge trannies) you make the primary
magnetizing inductance much more tolerant of unbalance. Unfortunately,
you also reduce that inductance by quite a bit! If you were using
triodes (low plate impedance) you might get away with it. For pentode
push pull outputs, you'd be wasting a good transformer by gapping it.
Pretty well all of the effects of fooling around with the air gap will
be in the very low frequencies where the effects of primary
magnetizing inductance dominate (as I'm very sure you know!).
Biasing the PP tubes (DC) for minumum unbalance will optimize
performance for low level, low frequency signals. Balancing gains on
the output tubes (not often done) will allow you to get more perfect
elimination of the even harmonics - something that has kinda fallen
out of favour.

> There is a spot a "reverse knee" might we
>say at the bottom of the BH curve but the use of an air gap by itself
>doesn't get us over this knee per se...might actually make it more
>difficult to overcome...by forestalling it by the demand for more
>exciting force to be applied since with an air gap some of the
>exciting force is expended in the air gap itself. Partridge
>recognizes this explicitly.

.....[snip!]....

I've tried to measure the effect of the little knee at the bottom of
the B-H curve. I never had any luck. I eventually assumed that you
could only do this with a very uniformly shaped material, that has ALL
had the same magnetic history. In a typical tranformer there might be
parts of the magnetic circuit where this is happening. I suspect it's
darn near impossible to bias the entire core with the same
magneto-motive force to see effects from such a narrow part of the the
B_H curve.
What amazed me while fiddling around with this stuff, is just how
much 'memory' a standard PP transformer has. It was damned annoying
trying to measure the magnetising inductance. After a few milliamps of
unbalance, and off it would go - variations of up to 30-200% on many
transformers. Usually a few good smacks with a plastic hammer would
bring the inductance back!

>Regarding nickels...yes they have much, much higher initial perms than
>the silicon steels. But the transformer gods declared that there
>would be no free lunch. So the fly in the free soup is that the
>nickel materials saturate very early vis-a-vis the "ordinary" grades
>of steel like the M6 material. It's a trade at best...
>

The Partridge trannies I have seem not to have the very high initial
permeability you mention. They showed a much better behaved linearity
than most of the modern steels (which I assume are grain oriented for
any of the better transformers). I doubt that this was much of an
advantage, because it's usually at the cost of permeability. I don't
know if they didn't use grain oriented steel then, or maybe the
technology of magnetic steels wasn't up to the level it is now.
As you're in the industry - what is the history of the various
steels? I imaging there must have been quite a flurry of development
back when tubes were king.

>I very much respect the work Partridge did and he was very thorough
>for the most part. His strategy works but again with an expense of
>increasing the exciting current which itself limits the ability of the
>iron to deliver clean power at low frequency signals. I have a paper
>by Ercel B. Harrison on this. If you would like a copy please let me
>know.

I'd be curious to know if the 'clean power at low frequency signals'
just applied to low level signals, or if he got decent performance at,
say, half to full power. I suspect that anything that makes
performance dependent on bias, usually means there is some very
non-linear mechanism going on. Personally, I'd sooner have some low
frequency loss than that distortion you get with magnetizing
inductance saturation. It means shittier specs (low primary
inductance) but a cleaner sound. I guess marketing takes over. I had a
tranny that would give me down to a few Hz at low levels. At half
power it would poop out at 30-40 Hz, so any excessive bass would sound
just awful. Oddly enough, in all my PP amps that usually made the bass
more 'flubby' sounding. On my SE amp, the same effect seems to make
the bass sound much harder than it should. It might not be wise to
generalise about it though....

-Paul

[Michael S. LaFevre]

Paul Guy wrote:

> By having an air gap in a PP transformer (I've done this with a
> hacksaw to a nice pair of Partridge trannies) you make the primary
> magnetizing inductance much more tolerant of unbalance.

Wince! Oucho for the poor Partridges :=). But your right about the
effect noted above.

> Unfortunately,
> you also reduce that inductance by quite a bit! If you were using
> triodes (low plate impedance) you might get away with it. For pentode
> push pull outputs, you'd be wasting a good transformer by gapping it.
> Pretty well all of the effects of fooling around with the air gap will
> be in the very low frequencies where the effects of primary
> magnetizing inductance dominate (as I'm very sure you know!).

As regards the reduction of primary inductance when an air gap is
used...again your absolutely right...and not by a small amount as
you've also indicated. I have in the archives some empirical data on
several Freed tranneys that were tested for pri L using several
different stacking methods...e.g., 1X1, 3X3, 10X10, butt stack and
etc. the more effective air gap (stacking in larger groups even
alternately produces larger and larger eff air gaps) the less the
inductance and by substantial amounts.

If we simplify a bit and only look at attenuation of frequency...then
low plate resistance triodes will need vis-a-vis pentodes (again
generalizing a bit) far less pri L for maintenance of sufficient pri L
to simply maintain a given response at a specific low freq cut-off
point. But this ignores phase response and iron induced distortion
and for these reasons one may choose to design a trans with
significantly more pri L than what is needed simply for maintenance of
the low freq cut-off point.

You mention that it would be a waste to use an air gapped trans with
pentodes in PP. Partridge in his 1939 papers argues for use of air
gapped tranneys even in these applications for principally two
different reasons. First is to "flatten" (forgive my inelegant way of
stating this) the inductance vs. magnetizing force behavior and
secondly Partridge argues that the air gap reduces core induced
harmonic distortions even in a PP trans. In the near future I along
with Frank Duetschman hope to be able to replicate some of the testing
that Partridge did (Partridge also describes the testing methodologies
that he employed). So the jury is out...at least in this neck of the
woods.

> Biasing the PP tubes (DC) for minumum unbalance will optimize
> performance for low level, low frequency signals. Balancing gains on
> the output tubes (not often done) will allow you to get more perfect
> elimination of the even harmonics - something that has kinda fallen
> out of favour.

Again, agreed. My sense is that in push-pull anything you can do to
optimize the balance btwn the two sides (push and pull) would have to
be for the good. Whether or not the elimination of even harmonics is
desirable...this is a normative question...

> I've tried to measure the effect of the little knee at the bottom of
> the B-H curve. I never had any luck. I eventually assumed that you
> could only do this with a very uniformly shaped material, that has ALL
> had the same magnetic history. In a typical tranformer there might be
> parts of the magnetic circuit where this is happening. I suspect it's
> darn near impossible to bias the entire core with the same
> magneto-motive force to see effects from such a narrow part of the the
> B_H curve.

Great point. The uniformity of the magnetizing force throughout the
iron path is anything but uniform for most of your typical shaped
lams...and in some cases this may actually be a hidden benefit.

> What amazed me while fiddling around with this stuff, is just how
> much 'memory' a standard PP transformer has. It was damned annoying
> trying to measure the magnetising inductance. After a few milliamps of
> unbalance, and off it would go - variations of up to 30-200% on many
> transformers. Usually a few good smacks with a plastic hammer would
> bring the inductance back!

On many PP transformers more than a few milliamps of unbal plate
current is enough to start to get them into trouble. Have you
calculated the DC flux for a PP trans...say a ST70 output...with 10 or
20 or 30 mils unbalanced plate current? So depending on what
constituted the amount of unbal plate current you may have been
saturating the core. Also the eff perm with AC and DC combined in the
core will be much, much lower than if you keep DC out altogether. All
of what your say makes good sense.

> >Regarding nickels...yes they have much, much higher initial perms than
> >the silicon steels. But the transformer gods declared that there
> >would be no free lunch. So the fly in the free soup is that the
> >nickel materials saturate very early vis-a-vis the "ordinary" grades
> >of steel like the M6 material. It's a trade at best...

> The Partridge trannies I have seem not to have the very high initial
> permeability you mention. They showed a much better behaved linearity
> than most of the modern steels (which I assume are grain oriented for
> any of the better transformers). I doubt that this was much of an
> advantage, because it's usually at the cost of permeability. I don't
> know if they didn't use grain oriented steel then, or maybe the
> technology of magnetic steels wasn't up to the level it is now.

Depends on the vintage of iron that you have. Also note that Dr.
Partridge died in the middle forties...so that his company survived
him but he was not at the helm of the enterprise any longer. M6
(grain oriented silicon steel) did not become much available
commercially until the early fifties. My hunch would be that the
Partridge's did not use any of the high perm mats (nickel, permendur,
etc) in large signal output tranneys. Most likely they might have
used (pre M6 days) a material called M15 which has the same silicon
content as M6 but is not grain oriented and was most likely hot rolled
as opposed to cold rolled. M15 has vis-a-vis M6 exactly the
properties you describe as regards i.e., good linearity from an
inductance point of view as well as less perm and in-the-end less
ulitimate ability to produce large L's for a given size core with a
given number of turns.

> As you're in the industry - what is the history of the various
> steels? I imaging there must have been quite a flurry of development
> back when tubes were king.

It was wild times in the thirties and forties...dozens of shapes
competing for the designer's attentions....companies large enough to
develop proprietary shapes to stave off the competition. Discoveries
and refinements of electrical steels. I am hopeful someday of doing
more articles on magnetics for say Sound Practices magazine and this
was one of the subjects I would like to cover because it is so
fascinating.

> I'd be curious to know if the 'clean power at low frequency signals'
> just applied to low level signals, or if he got decent performance at,
> say, half to full power. I suspect that anything that makes
> performance dependent on bias, usually means there is some very
> non-linear mechanism going on. Personally, I'd sooner have some low
> frequency loss than that distortion you get with magnetizing
> inductance saturation.

Paul, again complete agreement with your sense of avoiding core
saturation. And so frequently today the magnetic components are driven
just as hard as the circuit designers drive tubes...close to the blow
up mode. But Partridge wasn't playing this tune. Drop me a private
email and I'll shoot you a copy of his article.

> It means shittier specs (low primary
> inductance) but a cleaner sound. I guess marketing takes over. I had a
> tranny that would give me down to a few Hz at low levels. At half
> power it would poop out at 30-40 Hz, so any excessive bass would sound
> just awful.

Hard for me to say on this count. If it were an entry level
transformer and clearly stated that the power rating was say 40 watts
at 30 (or 40) hertz then you got what was promised. Yes at small
signal levels the extension in the bass may have been greater...this
being the "product" of perhaps sufficient pri L to make such possible.
But in addition to adequate primary L you also need to keep the flux
density level below the knee of the curve...at 30 or 40 watts...the
trans may have run out of magnetic head room. So you need both "power
handling capacity" and sufficient "L" in order to deliver power at low
frequencies.

Great points you made Paul...I enjoyed it very much. Nice to have a
discourse on RAHE instead of a battle!!!

Mike

[Scott Frankland]

Paul Guy wrote:
>
> Scott must write articles for scientific journals -

Not yet. I wanted to try this stuff out on you guys first! <he said,
ducking!>.

> >The worst case SID occurs in class B amplifiers because here the
> >interwinding capacitance must be ramped up from empty (since class B
> >operates the tubes at zero idle current). This is consistent with the
> >charge integral equivalent of work
>

> Now here's the rub. If one side of the winding has zero current (as
> with class AB or B circuits, WHAT ABOUT THE OTHER SIDE? Remember the
> transformer windings in a PP transformer are COUPLED together.

Yes they are. I didn't consider the effect of mutual coupling.
Therefore this particular equation is invalid for PP transformers.
Evidently the total interwinding capacitance appears across the
COMPOSITE tube, rather than across each tube independently. This does
not necessarily invalidate my argument, however, since, when one tube
is near peak the other is near trough. In class B amps the rp is
substantially higher near trough than is the class A amp.

> Quite apart from all the formulae, my own experiences with SE and PP
> trannies is that you want leakage inductance and capacitance reduced,
> the compromise comes from the resonant effects, and plate and load
> impedances at high frequencies.

I was looking at the two kinds of load reactance considered
separately, and trying to show that they were BOTH related (to one
degree or another) to the class of operation (class A vs. class B has
already been shown to substantially affect the susceptibility to notch
distortion by the references I cited).

> If you measure an UNLOADED plate
> transformer impedance vs freq., the impdedance is inductive (and
> impedance linearily increasing) until a few kiloherz, and then
> capacitive (impedance linearily decreasing) with higher frequencies.

Right. And the capacitive load becomes progressively harder to drive
with frequency. That's why I advocated mininimizing it above all. This
is not to say that leakage inductance can be neglected. No. But in
winding a transformer, a designer has a choice of whether to optimize
for lowest interwinding C, or lowest leakage L. I reasoned that, since
the power sensitivity of a tube amp to an increasing load (as by
increasing L) is more tolerant than to a decreasing load (as by
increasing C), that it would be most beneficial to minimize shunt
C. That is the crux of my argument, whether or not I can find the
supporting theory.

> Probably more to the point, it's very difficult to run class B. A poor
> implementation will run marginal class C, where a small chunk of the
> waveform is lost. This is not an issue with class AB, and only of
> interest to those very few that are running good class B (like you can
> do with transistors and get very low distortion).

I am here making a distinction, not so much between class B (which is
hardly ever seen), but rather, between class AB and ENRICHED class AB.

> >The solution to notch distortion is exactl`y the same as that for

> >excessive interwinding capacitance: (1) reduce the leakage inductance
> >by skillful transformer design; and/or (2) increase the idle current
> >so as to enrich the class AB operation.
>
> The McIntosh amps are an example of (1). These were run class B, but
> utilized transformers that had very low leakage inductance (I believe
> they were wound trifilar or multi-filar, with the 2 plate windings,
> the output winding, and feedback windings - global and screen)

Frank McIntosh pointed directly at the problem of quasi transients
(what he dubbed "notch distortion") as blocking the goal of higher
efficiency in audio amplifiers. His approach was to minimize leakage
inductance, since this is the path taken by the quasi transient. In so
doing, however, he also reduced interwinding C drastically (by
reducing the number of turns). This is of course the most preferable
route to transformer optimization.

> >push-pull amplifier will have less total distortion at the top-end
> >than will a single-ended amp, in spite of the fact that single-ended
> >amps necessarily operate in class A. That is because in most cases
> >there will be less leakage reactance in a push-pull transformer [7].
>
> Naah, that's just too general, and you've left too many loopholes to
> make a good argument. I have a SE tube amp running right here that has
> very low distortion!

But I said, "in most cases". I still think that's true.

> Until you run a PP amp right into class B, or overdrive the SE stage,
> the direct affects of winding capacity and inductance aren't a real
> cause of distortion.

True, but the slew-rate limit is 2pi x fmax x Vp. In other words, it
DOES occur at or near the peak voltage. And that's what I'm concerned
about. The slewing does not always occur right at the limit,
however. In a zero-feedback amp, there will be a graduated approach to
the slew-rate limit, whereby the amp will begin to soft slew at
perhaps 2/3 of the limit.

> Even the effects you claim are happening, how significant are they? I
> have yet to see them. Certainly they are not in the top 20 concerns
> when optimizing a tube amp.

I was looking for a theory that would explain what I have measured.
Specifically, if an amp is tested for THD at 10kc vs. 1kc, I found
that the ratio is smaller for the class A amp than for the class AB
amp. Thus, the amount of idle current has a direct bearing on this
ratio. The only parameter I can think of that might explain the
difference is capacitance (and the attendant slew-rate limiting that
accompanies a capacitive load). One other possible explanation is eddy
current loss, which also increases with frequency.

> Maybe from a frequency response, or phase
> margin point of view, but distortion? Do you have numbers? This should
> be an easy experiment to verify, just add some parasitic capacitance
> via discrete caps hung off the primary. Then measure high and low
> level distortion. Please don't make this another one of those horrible
> audio myths!

I'll try this when I get back from CES and report back to the group.
I certainly don't wish to perpetuate yet another audio myth (there are
too many of those already)!!!

~SF~

[Scott Frankland]

Michael S. LaFevre wrote:
>
> secondly Partridge argues that the air gap reduces core induced
> harmonic distortions even in a PP trans.

This might explain the increase in distortion at high frequencies due
to eddy current loss, since, with a gapped tranny there will be a
reduction in such currents.

~SF~

[Scott Frankland]

Michael S. LaFevre wrote:
>
> So independently of whether the trans could be designed to
> handle unbal DC plate currents (we know that they can at a price) is
> this an appropiate strategy?

The idea I'm attacking is that PP amps cannot be made to produce
even-order distortion products. I'm not suggesting that unbalancing
the DC bias is the optimal way to do this, it's just a way (a poor
way, in fact). A much better way would be to provide unbalanced
feedback around a balanced amplifier. I've done this with diff-amps in
line stages (experimentally) and found that it does indeed create a
distortion spectrum much like that of an SE amp (and it's tunable to
boot!). If people are going to haggle about distortion products (such
as Jean Hiraga seems to do), then it might as well be known that PP
can do it pretty much as well as SE. But is this really the secret to
the single-ended sound?

> There is a spot a "reverse knee" might we
> say at the bottom of the BH curve but the use of an air gap by itself
> doesn't get us over this knee per se...might actually make it more
> difficult to overcome...by forestalling it by the demand for more
> exciting force to be applied since with an air gap some of the
> exciting force is expended in the air gap itself.

Right. That's why I suggested the idea of unbalancing the DC
currents. Of course, this brings other troubles. But what are these
troubles compared to those of an SE? Not that much worse, perhaps,
just different. In any case, I'm not actually advocating this
practice, just suggesting ways in which even-order distortion can be
generated in PP amps.

> Regarding nickels...yes they have much, much higher initial perms than
> the silicon steels. But the transformer gods declared that there
> would be no free lunch. So the fly in the free soup is that the
> nickel materials saturate very early vis-a-vis the "ordinary" grades
> of steel like the M6 material. It's a trade at best...

I realize this. What in audio is NOT a trade? :-)

~SF~

[Paul Guy]

More techno-babble about PP and SE output trannies! This is getting
into a fairly serious engineering discussion, something I'm not sure
is really welcomed in this newsgroup.... Anyhow, here goes:

[ Unless you've got your tongue firmly in your cheek, you must have
mistaken us for someone else. One of the very best things about
r.a.h-e is that it often examines audio with a critical, technically
informed eye. Given the amount of absolute rubish being spewed at the
public by the audio industry (manufacturers, dealers, and press), this
is a sorely needed breath of fresh air. Yes Paul, it's welcomed. Of
course, purely aesthetic discussions are welcomed too. -- jwd ]

Scott Frankland <audi...@ix.netcom.com> wrote:

>> If you measure an UNLOADED plate
>> transformer impedance vs freq., the impdedance is inductive (and
>> impedance linearily increasing) until a few kiloherz, and then
>> capacitive (impedance linearily decreasing) with higher frequencies.

>Right. And the capacitive load becomes progressively harder to drive
>with frequency. That's why I advocated mininimizing it above all. This
>is not to say that leakage inductance can be neglected. No. But in
>winding a transformer, a designer has a choice of whether to optimize
>for lowest interwinding C, or lowest leakage L. I reasoned that, since
>the power sensitivity of a tube amp to an increasing load (as by
>increasing L) is more tolerant than to a decreasing load (as by
>increasing C), that it would be most beneficial to minimize shunt
>C. That is the crux of my argument, whether or not I can find the
>supporting theory.

That dependency on load variation depends a lot on whether you are
running pentode or tiode (or even ultra-linear), and how the load line
compares to the optimum for power effeciency.

At the frequency extremes, the transformer tends to give you a low
impedance. That can be adjusted so for example, the reactance has a
value thats comparable to the nominal load impedance at 20HZ and
20KHz. Between those frequencies, it rises up to a peak somewhere in
the midrange, where the reactance might be a hundred times the nominal
impedance.

Knowing the plate resistance, you could calculate the attenuation
behaviour. There would be obvious saturation effects at the very low
frequencies that would give rise to some obvious bass distortion. At
the high frequencies that effect doesn't exist, the tube just gets
shunted out. Harmonics aren't a serious problem - most of them are out
of audibility, and the series leakage inductance its various
interactions usually attenuate them anyhow. Given that you must
compromise, good sound would dictate that you'd throw away high
frequency performance first UNLESS you intend to use a lot of
feedback.

>Frank McIntosh pointed directly at the problem of quasi transients
>(what he dubbed "notch distortion") as blocking the goal of higher
>efficiency in audio amplifiers. His approach was to minimize leakage
>inductance, since this is the path taken by the quasi transient. In so
>doing, however, he also reduced interwinding C drastically (by
>reducing the number of turns). This is of course the most preferable
>route to transformer optimization.

....That would be an interesting design to see. Can't speak much for
audio bifilar wound coils, but otherwise bifilar windings have a real
problem with high coil-coil capacitance (actually most designs take
advantage of it). If you think how a bifilar coil is wound (two wires
wrapped as a pair on the coil form), there has to be a much larger
capacitance between them than any other construction technique.

However, I think that might not be a big problem with a PP amp, it
probably is more at the mercy of capacitance to ground (or case, or
secondary). Maybe by putting both windings in one, and reducing the
coil surface area (especially at the 'hot' plate side, McIntosh was
able to reduce that 'stray' capacitance, at the expense of coil-coil
capacity.

>True, but the slew-rate limit is 2pi x fmax x Vp. In other words, it
>DOES occur at or near the peak voltage. And that's what I'm concerned
>about. The slewing does not always occur right at the limit,
>however. In a zero-feedback amp, there will be a graduated approach to
>the slew-rate limit, whereby the amp will begin to soft slew at
>perhaps 2/3 of the limit.

Are you really talking about slewing as is used in control theory, and
most feedback amplifiers, or just the ramping up and down that you get
driving a capacitive load (not under overload conditions)? Slewing
implies fairly gross non-linear behaviour, at all points of the output
waveform. Simply driving a square wave into a capacitive load does not
cause slewing,providing that the output current is still controlled by
the input signal. Once the output current is independent of the input
signal, then you're in trouble, and slewing may happen. The real test
for slewing vs. ramping or whatever other names you choose, is this:
if you put in another (lo freq) signal with your 'slewing' signal,
does it appear on the output? If the other signal shows up (it should
if things are linearly driving a capacitive load) then you don't have
slew rate limiting.

I guess I'm used to seeing slew rate limiting show up where you
wouldn't expect it. Usually it becomes a problem when apparent gains
and outputs are such that you know (or think you do) that there is
going to be no clipping or overloading. Then, because of feedback and
limitations of internal stages, some driver stage gets pushed into
overload because it can't drive an internal capacitance hard enough to
keep up the input signal. That creates a big error signal which just
makes things worse. The output signal is completely at the mercy of
the offending stages ability to drive that capacitor, and is virtually
unaffected by any input variations.

>> Even the effects you claim are happening, how significant are they? I
>> have yet to see them. Certainly they are not in the top 20 concerns
>> when optimizing a tube amp.

>I was looking for a theory that would explain what I have measured.
>Specifically, if an amp is tested for THD at 10kc vs. 1kc, I found
>that the ratio is smaller for the class A amp than for the class AB
>amp. Thus, the amount of idle current has a direct bearing on this
>ratio. The only parameter I can think of that might explain the
>difference is capacitance (and the attendant slew-rate limiting that
>accompanies a capacitive load). One other possible explanation is eddy
>current loss, which also increases with frequency.

A PP amp effectively doubles the transconductance of a single tube
while in class A. In class AB, there are points where one tube is
cutoff, then the overall transconductance drops to the single value.
There would be a 'kink' in the composite characteristics of the PP
tube stage. There is an opposite problem to this that you'll see in
class B amps. There you avoid biasing them both on 'cause the
transconductance will double during that overlap. This is one of the
reasons that biasing transistor amps can be fiddly.

Where the plate resistance is important, things can be affected when
one tube is cut off. From the output load (and all the parasitic
inductances, capacitances) point of view, the driving impedance will
always be half the plate resistance in class A, but will rise to the
single plate impedance in class AB during large signals.

Those might be a reason for your measured increased distortion at high
frequencies in class AB. I'm assuming that you're running the tubes
fixed bias (adjusted for the appropriate class) and that the
amplitudes of the test frequencies were significantly below clipping.

...hmmm now that I've though of the problem of transconductance
halving (or doubling) it puzzles me why this wouldn't be a bigger
issue. I've never heard much of it in the tube world. Maybe because of
the fact that it only kicks in at quite high levels no one really
cared about it. The other possibility is that its distortion would not
be as obnoxious as clipping, or that the transconductance in a real
tube is all over the map anyhow, especially at the output signal
limits. Another is that transconductance effects are more likely to
show up with low impedance loads. Another is that transconductance
varies a lot with bias (signal), and when one tube cuts off, the
other's transconductance was getting larger anyhow, so as to minimize
the overall effect except for a bit of a 'kink' in the curve.

Maybe a new thread for the tube fiends?

-Paul