euphonic distortion

[kelly]

This one goes out to all our tube lovers...

What exactly is euphonic distortion?

When you find it in an amp or source, do you like it with all instruments? All kinds of music? Or only some?

 

[carlo]

Euphonic distortion - To soften and sweeten the sound of components in an attempt to mistake warmth and excessive bloat for realism.

One of the interesting things about tubed components is that one can try to get distortion and a forgiving sound should they choose to: tubes on their last legs tend to give a loose disjointed sound that, while far from any semblence of what the recording is giving, also pleases and relaxes. Are tubed equipment inherently soft and mushy sounding? No. But most can be made to sound that way with enough effort.

As a generality part of the second order harmonic distortion and the tube effect, in my experience, leads to a layering of notes and a sense of decay. Once again, this is obviously distortion, but is that a bad thing? The sense of realness and soundstage offered by tubes, as a generality, is part of their appeal. At some point distortion turns into notes taking shape and getting life... the line between it all gets narrow for some and is as clear as day for others.

For this listener euphonics are for acoustics, brass, and winds. All benefit from some softening but exaggeration on body, all are the victims of bad recording quality (more so than electrics in my opinion), all need something that distortion, in its own way, provides.

The duality of this is, it has to be taken with a goal and purpose in mind. I regularly run soft sounding tubes through a AH! Njoe Tjoeb 4000 with the explicit intent to soften badly recorded CDs (regardless of genre or primary instrument). It makes something bad sound better by messing it up some - I'm comfortable with that. Could I live with an amp or source with the above distortion as a primary part of my system? No, and I've tried. But it offers something and lets me listen to somethings that I just can't listen to otherwise.

2 pennies and all that,
carlo.

 

[AndreYew]

Another kind of euphonic distortion (good-sounding distortion) are imaging artifacts, such as those found in LPs. Because of the way LPs work, there are several interchannel distortion mechanisms that make things like imaging and soundstaging sound better, so euphonic distortion isn't only about frequency response bumps and dips.

BTW, I am not trying to cast a perjorative light on LPs or euphonic distortion --- we should listen to what we like and enjoy it. Stereo is already intractably inaccurate, and some euphonic distortion won't spoil its assumed purity.

--Andre

 

[carlo]

Andre,

Well said. Your post got me to thinking: for whatever reason I like "euphonic distortion" to happen at the source, not at the amplification stage. I feel there's more control of it there, a tubed CD player (output stage or, to less of a degree buffered) can have outstanding control of the frequency response while still providing the soundstage that lets you know there's a tube there. I don't know why it is, but with amps the likelyhood of that effect being pleasing (as opposed to irritating) is much smaller.

To flip the idea around a tad, the things that I've associated with even-order harmonic distortion (depth of decay, sense of space, soundstage) is what I got from upsampling through a Cary 306/200. Interelated I'm not qualified to say, but the sound characteristics were similar.

2 more pennies.
carlo.

 

[AndreYew]

Carlo,

Interesting thoughts and observation. Perhaps it's better to introduce distortion at the source, and keep the rest of the listening chain relatively linear so that the interaction between components is well-controlled and predictable, so you can dial in the kind of sound you want. If every element of your chain were non-linear, they'd interact in basically factorial combination --- each element of the chain would have its own set of interactions with every other element of the chain.

Mathematically, if something has 2nd order distortion, its transfer function looks something like:

t(x) = A1*x + B1*x^2

A1 is the gain of your system
B1 is the amount of 2nd harmonic distortion

Let's say you have another component downstream that has the same transfer function, and let's call it u(x) instead. Using both components in series (perhaps t(x) is your preamp, and u(x) is your power amp), you basically perform the following operation:

u(t(x)) = A2*t(x) + B2*t(x)^2 (A2 and B2 are similar in function to A1 and B1)

Expanding this out, you get a much more complicated equation:

u(t(x)) = A2*A1*x + A2*B1*x^2 + B2*A1*A1*x^2 + 2*B2*A1*B1*x^3 + B2*B1*B1*x^4

Rewriting all the constants to simplify the expression, you get:

u(t(x)) = A*x + B*x^2 + C*x^3 + D*x^4

A is the overall gain of the two components
B is the amount of 2nd order harmonic
C is the amount of 3rd order harmonic
D is the amount of 4th order harmonic

Some interesting things to note:

1. We now have harmonics that weren't there before: the general rule is that components cascaded together will make harmonics whose order is the sum of the highest harmonics of each component separately. So if one of the components had 3rd order distortion, and the other 2nd order distortion, we would have 5th order distortion when those components are used together.

BTW, feedback and cascaded stages will compound harmonics within a component, which is why Kevin Gilmore in his Rules of Proper Audio Design says that individual stages within an amplifier must already have linear behavior without feedback, and why many designers like Nelson Pass prefer amps with fewer stages.

2. For simplicity's sake, my calculation assumed our components have only 2nd order components. In reality, real components have higher order components, as you can see in any Stereophile review with measurements. So the situation is way more complicated than my simple example.

3. Also for simplicity's sake, I only used two components. A real system often has more, and the complexity compounds.

4. Why does an nth order polynomial term imply an nth order distortion? The easiest way to see this is to assume x = sin(t), and expand out the terms. You'll see that one of the terms will have sin(n*t), which is an nth order harmonic.

So if we have one non-linear component, we can control fairly well what euphonic distortion we want. When you start using a bunch of them, they interact in very complicated and wide-ranging ways.

--Andre

 

[carlo]

Andre,

Which, in a sense, is what I guess system matching is all about. VAC has an interesting piece on their website that suggests higher odd numbered harmonics, while lower in measured distortion, is grating to the human ear.

Many of the amps I enjoy have no feedback - there's something that, in light of the fact that we're talking about distortion, sounds truer to the source to me. The line between what can be measured and what can be heard gets fuzzier as time passes, as does the search for "accuracy." There's no generalities or absolutes I can draw from that, its just the way it is.

Great post, its been years since I looked at anything even resembling applied math and your explanations helped a lot

best,
carlo.

 

[AndreYew]

Carlo,

Thanks. What VAC says is true, and one reason is because we can hear higher harmonics (up to our hearing limits of course) better because they are farther away from the fundamental, and aren't masked, or easily related to it.

--Andre

 

[kelly]

I'm afraid this conversation quickly lept over my head and there would be too many math prerequisites to ask for explanation of the technical stuff. Sorry.

Let me ask this, though:

Many people are fond of the "tube" sound and different people have an idea of what exactly that tube sound is. Do you guys think the tube sound is achievable with solid state components?

I do think some tube equipment manages to not sound so "tubey." I do believe that some people choose to use tubes in their circuits simply to avoid some of the distortion types caused by solid state components (and as a consumer hope some day this will go away when solid state components get better).

This has been a difficult concept for me to grasp. It has honestly only been within the last year that I've come to realize that some audio designers intentionally add distortions to their components to "sweeten" the sound. I expected this in the pro audio and musisian's world where, for example, someone might want a particular electric guitar to sound a certain way. I'd have never expected that anyone would want such a strong signature attached to every instrument in an orchestra so I still struggle with this idea.

 

[Tomcat]

Quote:

As a generality part of the second order harmonic distortion and the tube effect, in my experience, leads to a layering of notes and a sense of decay. Once again, this is obviously distortion, but is that a bad thing?

carlo,

I don't belive that distortion is the resaon for superior musicality. I haven't studied the etymology of the term "euphonic distortion" but I bet what gave rise to it is the fact that, when solid state amps were introduced in the late sixties, early seventies (?), they measured beautifully but sounded horrendous compared to tube equipment. Hence the idea that the fairly high amounts of second order harmonic distortion that can be measured in a lot of tube amps are the reason for their euphonic and more musical sonic signature. What to do if a psychoacoustic theory doesn't conform with human perception? Exactly: let's declare human perception flawed.

It seems to me that there would be a much simpler explanation for the musical superiority of tube amps: they are better. They are more accurate in areas that are far more relevant to a human listener than THD figures. I vividly remember the sound of Accuphase amps from the late seventies/early eighties. Those were about the most constipated, intransparent, uninvolving and unmusical amps I have ever listened to. And they measured great, of course. They had THD figures with four decimal places. And they had a large dose of global negative feedback in order to achieve this. But what for? It's just as you have said:
Quote:

Many of the amps I enjoy have no feedback - there's something that, in light of the fact that we're talking about distortion, sounds truer to the source to me. The line between what can be measured and what can be heard gets fuzzier as time passes, as does the search for "accuracy."

"As time passes" you say. That's a point stressed by Audio Note, a British manufacturer (famous mostly for their tube amps): Accuracy in the time domain is of far greater importance than low THD figures. But what I believe is most important concerning euphonic distortion: listening to the quality of the music reproduced by a system is much more relevant than conducting measurements in order to support some half-baked psychoacoustic theory.

 

[bootman]

Quote:

Originally posted by kelly Do you guys think the tube sound is achievable with solid state components?

Some have tried.
Bob Carver and his "T-mod" amps come to mind.

Here is an interview with Carver that touched on some of this.

 

[AndreYew]

Tomcat,

Better does not necessarily mean accurate. People who enjoy tube amps should simply enjoy them, and not try to idealogize their choice. Same thing with solid state amps.

BTW, THD is almost completely worthless as a measure of any audio equipment. Two amps with identical THD to four figures can sound completely different.

--Andre

 

[Anders]

kelly,
I think there are two aspects of euphonic distortion to consider.

The first is to create a specific sound that not gives a neutral reproduction and in some sense enhances the sound. I don't argue against that if people like it and if they don't critisise more realistic systems as bad, because they don't are euphonic in the way they want. My experience is that this can sound terrific on some albums or tracks but awful on others. This approach could be more useful is you only listen to a specific type of music as jazz, chamber music or heavy metal.

The second is a more practical way to optimise sound. The source material (e.g. CDs) are distorted and the system adds further distortion. In some cases, this distortion can be made to sound more agreeable by adding more distorion of a benign type, e.g. a very detailed and hard sounding system can become more pleasing by a slight decrease in resolution. Maybe, something similar happens in vinyl playback by channel crosstalk and the addition of some hiss.

I have not heard them but sometimes see products from Audio Analogue praised for the vinyl-like, analogue sound (CD players and amplifiers).

Carlo and Andre,
Your discussion is very interesting and I believe that distorting the source component could be a good idea. But wouldn't the best be to have two CD players, one for "good" recordings and one more forgiving for "bad" recordings?

 

[kelly]

Anders
My CD player has been modified with a tube output stage and I think many would describe it as a more analogue-like sound, but it is not because it is lower in resolution. Or rather, maybe it is lower in resolution than what the complete mods would have been if a tube had not been used in the output stage but even that, I suspect is incorrect.

I really think there are two things that tube output stage on a CD player can accomplish:
1. It can substitute one kind of distortion for another (see all tube vs ss arguments). Taking away the ss type of distortion can be to an overall benefit depending on the design and implementation.
2. It can add a coloration that is preferable for some people on some recordsings/some genres, etc. as you alluded to.

I agree with most of what you're saying but I'm not sure where resolution fits in. I have heard some tube amps that sound slightly blurred compared to other amps but even then, I don't know that a tube amp could not be contructed to have a tighter focus and better seperation.

 

[carlo]

kelly,
Quote:

different people have an idea of what exactly that tube sound is. Do you guys think the tube sound is achievable with solid state components?

bootman hit this point well - what many generalize as a tube sound (from my perspective this is a fat, bloatish coloration) can be acheived through transistors given enough effort. whether it sounds like a tubed component is arguable, as is the idea that tubes inherently have certain sound characteristics. coloration for the sake of coloration is what most of us try to avoid, but the fact remains that people pay good money to hear second order or euphonic distortion.

Tomcat,

exactly. when i first entered high end audio my search was for the most accurate, transparent components available and numbers pertaining to watts and distortion were highly relevant to me. in the process of gaining experience and learning what sonic characteristics i'm looking for (call it personal bias) i realized that "accurate" and "neutral" are words thrown around carelessly. as andre (very well) stated there isn't much point in treating components and the signal as a sacred entity - the point is to acheive the best sound one can from an already comprimised recording.

the acheivement of that goal involves measurement, trial, and evaluation - just not necessarily by a machine.

Anders,
Quote:

But wouldn't the best be to have two CD players, one for "good" recordings and one more forgiving for "bad" recordings?

thats certainly my perspective, but its not one shared by my peers and its of course dependent on finances (i had to give up my modified arcam alpha 9 to cover rent, now i'm stuck with a cd player i only kept for the explicit purpose of enjoying bad recordings).

players/sources without blatant softening of the sound, ones that get closer to what the disc is giving without masking, should certainly be the primary source for the discerning audiophile. however, there's a lot of great music made by people who didn't have access to great recording equipment (the animals, the kinks, louis armstrong, the velvet underground's true garage stuff, etc) that, well, sound ungodly bad through that equipment. i rationalize the fuzzing out and sweetening of those recordings (once again, through a tube buffered cd player utilizing forgiving and soft sounding tubes) with my love for the music.

my goal with high end audio is to get closer to the music, but there is a balance between what can be measured and what can be heard. finding it is the hard part.

best,
carlo.

 

[Anders]

Kelly
I wrote from the perspective of aural experience. I didn't refer to to tube vs. SS. It is true that there are tube components that correspond to a common mis-representation of tubes as overly soft with lacking transient response and bass. This is not necessary at all, but I suspect that there are still tube lovers that want this. And I do believe that there are tube systems that are as good as SS, but without the disadvantages that are common in the latter.
Maybe, my example of some correlation between resolution and harshness was not good. I have experienced instances of this in SS, but also the contrary when you get increased resolution without trade-offs. But if you have harsh sounding distortion, a little smoothing distortion can be beneficial.
I think my system would benefit from a tube modification of my CD player. I don't know if this is possible and it would incur an unreasonable cost. I wait and hope to find a good SACD player that also plays CDs with upsampling with better quality than I have now.