Stereophile | January 2017 Issue

[microstrip]

Just agreeing with those 1999 observations.
Still do. Nobody has yet produced a cogent reason why people buy these items other than this.

For me it is no problem to buy what you like to listen to best, that makes sense.
If you want to justify it to yourself any way you like, fine.

Unfortunately we are only debating with the static old fashioned measurements that many high-end designers consider insufficient to characterize amplifier subjective sound quality. Measurement technology evolved a lot - modern analyzers supply lots of information on dynamic and transient behavior, but we do not have information on recent developments and go on using the same old arguments, often discussed in other audio forums, including the DIY community that spends a large time on the subject. THD is insufficient, spectral analysis is a little better, but IMHO still insufficient to explain what we hear.

BTW, I was just checking the spectra of distortion of one amplifier I am selling to a good friend - comparing the performance of both channels is a great and reliable way of checking its condition. Besides, some tube amplifiers have an internal potentiometer that can be adjusted to minimize second harmonic - spectral analysis is really a great tool.

 

[f1eng]

So you have the Animas?Very nice. We got best of show with the excellent Animas in Munich's 2011 show paired with our full electronic set up.With our Exsequor amps.SE though

I remember that!

 

[LL21]

As most solid state amplifiers Gryphon's have plenty of feedback - they only claim zero global negative feedback.
We should not be afraid that it sounds different from SET's!

interesting and good to know. have no fear....I like what i hear with Gryphon...and have never had any desire to look elsewhere. That said, i admit i have not spent sufficient time with SETs (Lamm, Kondo) to really get to know them, and certainly have never compared in our system. i have heard Kondo in another system, and found that system enthralling. One interesting observation is that a number of the reviews i have read about Gryphon each seem to liken Gryphon's tonal purity specifically to that of SETs, while obviously have the sheer power to create endless reserves of dynamic range and bass slam.

 

[LL21]

So after watching the most welcome tennis match of ideas, science, calcs, i make the following observations:

1. All amps have some level of distortion...whose pattern and type various by design and implementation. measurements of every order, how much matters and does not matter or is inaudible: all is not a fully understood science to the nth degree...more work to be done
2. The human ear is not a machine and seems to have its own 'in take/processing science' which must be taken into account when designing an amp which is always going to have some level of distortion.
3. Amp design goals are to minimize distortion...but probably ALSO to minimize that distortion in a way that complements the human ear/hearing/perception mechanics
4. The concept of 'euphoric' distortion is probably a personal taste thing even if there are generalizations one might attempt to take. So care must be taken in HOW human ear/hearing/perception mechanics are incorporated into any design. Ideally, you focus not on simple human preferences but more on actual human physiological ability to hear/discern certain kinds of distortions
5. Adding in the overall system to amp design/choices, its clearly important to focus incredibly hard on the amp-speaker combination. An amp that is loafing along, easily able to perform at its best with lowest distortion when connected to a particular speaker is clearly a good place to start...and in an ideal world perhaps the amp has been specifically designed to drive the exact load characteristics shown by the actual speaker. (Here come the Active Speaker proponents!)

Basic Question: Is it easier to design a low distortion 10-watt amp than a 200-watt amp? Is it easier to design a low-distortion 100db efficient speaker than an 85-db speaker? This may be a ridiculous question, in which case I withdraw the question, your honor.

Otherwise, carrying on: if the speaker answer is "equally easy/difficult to design the efficiency", but the amp answer is "easier to design low distortion low-watt amp"...should that lead us to search our favorite of high efficiency speaker...and go for amps (whether higher powered or lower powered)...that are as low distortion as possible at 1-10 watts?

 

[opus112]

1. All amps have some level of distortion...whose pattern and type various by design and implementation. measurements of every order, how much matters and does not matter or is inaudible: all is not a fully understood science to the nth degree...more work to be done

What I've noticed is missing on this thread is the type of distortion. THD isn't the audible distortion present when listening to music - it only shows up on sinewave testing. The distortion which its possible to hear on music (when your system is sufficiently sub-standard) is intermodulation distortion. This is a large reason why THD measurements aren't telling us very much about amplifier sound. This important detail has been understood since the work of Brockbank and Waas many decades ago - they showed that IMD products would completely submerge THD harmonics when the stimulus is normal music.

 

[morricab]

What I've noticed is missing on this thread is the type of distortion. THD isn't the audible distortion present when listening to music - it only shows up on sinewave testing. The distortion which its possible to hear on music (when your system is sufficiently sub-standard) is intermodulation distortion. This is a large reason why THD measurements aren't telling us very much about amplifier sound. This important detail has been understood since the work of Brockbank and Waas many decades ago - they showed that IMD products would completely submerge THD harmonics when the stimulus is normal music.

Actually the one thing that was really NOT being discussed on this thread was THD. Geddes showed in a couple of paperd that there is NO correlation between THD and sound quality. He also showed no correlation between IMD and sound quality.

The harmonics generated, either pure or through intermodulation, will determine audibility and obnoxiousness of the distortion. Correct weighting is necessary for a good correlation with perception.

 

[opus112]

Geddes showed in a couple of paperd that there is NO correlation between THD and sound quality. He also showed no correlation between IMD and sound quality.

Got a link to this claim on IMD? I am familiar with some of Geddes' work and even entered into an attempt at questioning him about it on DIYA but he turned out to be fairly defensive. The paper he referred me to examined and modelled only static distortions which in my view renders part of it irrelevant.

 

[Robh3606]

Is it easier to design a low-distortion 100db efficient speaker than an 85-db speaker?

Hello LL21


I think because of the higher sensitivity the higher efficiency speaker will tend to be more linear and operating in that range more often than not. As I see it the higher efficiency has benefits all the way down the line as all components will be operating in essentially a low power mode at listening levels we would typically listen at. This is even more important if we listen at anything near life like levels as the 85db speaker is decidedly on the wrong side of the power curve and both the amplifier and individual drivers will be working quite hard at that point. Lets face it a 15db advantage is huge as far as the power levels required for 110dB peaks. Even at an average of 95db you are looking at 10 watts vs milliwatts That's at 1 meter and does not account for distance to the speakers. Distortion levels in higher efficiency drivers can be quite good so an advantage all around.

Attached is a distortion measurement on JBL KS9900 speaker system at 96db you can see that the distortion is under .3% in the midband and quite good over the entire bandwidth.

Rob

 

[morricab]

Got a link to this claim on IMD? I am familiar with some of Geddes' work and even entered into an attempt at questioning him about it on DIYA but he turned out to be fairly defensive. The paper he referred me to examined and modelled only static distortions which in my view renders part of it irrelevant.

Both papers can be found on Geddes website

 

[YashN]

SET amps produce lots of distortion

That statement itself is misguided.

but usually sound nice therefore the distortion they are producing is euphonic.

That is incorrect too: the distortion doesn't make things better: the 2nd order distortion doesn't sound as horrible as 3rd order.

There are some nuances there that appear to elude you.

The idea that there is some magic about SETs which makes them sound nice despite being non linear has made a renaissance, I don't buy it and I have never seen a convincing hypothesis as to why such magic could exist.

Simplicity of circuit is one, the restitution of transients is amazing.

I am still convinced that people like SETs because of their distortion characteristic is changing the sound to their taste. I see nothing wrong with making such a choice btw.

Flawed reasoning => Flawed deductions.

 

[YashN]

It may not be the goal, but it is always the result.

No: a good SET designer will work to reduce all distortions

This means that adding harmonics will definitely change the timbre and all SETs add harmonics since the basic circuit is not sufficiently linear to avoid doing so.

SS amps can have more 3rd order distortion, which is even more damaging to SQ...

I quite enjoy the sound of SET amps, but I know they are rose tinting the sound.

A good one would be as neutral as possible.

 

[YashN]

that SETs ... and most of them have an uneven frequency response due to poor speaker impedance matching.

Ah well, if you judge a SET by deliberately using it with the improper speakers, it is rather your faults as a system builder which should be criticised.

 

[f1eng]

That statement itself is misguided.



That is incorrect too: the distortion doesn't make things better: the 2nd order distortion doesn't sound as horrible as 3rd order.

There are some nuances there that appear to elude you.



Simplicity of circuit is one, the restitution of transients is amazing.



Flawed reasoning => Flawed deductions.

I agree that flawed reasoning leads to flawed deductions, but it is your reasoning which has no basis in fact.
Simplicity of circuit has no merit if it results in a non-linear solution, it is poor engineering understanding.
There is evidence, quoted by another poster, that people found the addition of 0.5% 2nd harmonic distortion sounded "more musical" than the undistorted signal. Daft but that is why some people say the same about SETs, the concept of the equipment rather than the musicians being musical is ridiculous.

You are just another person trying to justify your preference to yourself without admitting that it is the added distortion which you like the sound of. IMO that is the only explanation that supports the facts, all the rest is self delusion.

 

[f1eng]

Ah well, if you judge a SET by deliberately using it with the improper speakers, it is rather your faults as a system builder which should be criticised.

Ha-Ha-Ha!
Are there any speakers with a flat enough impedance characteristic to have its FR inaudibly changed by a SET amp,given that 0.1 dB change is audible?

 

[microstrip]

(...) You are just another person trying to justify your preference to yourself without admitting that it is the added distortion which you like the sound of. IMO that is the only explanation that supports the facts, all the rest is self delusion.

Surely we like the sound because something is added and something is not added. Your explanation does not add anything to the discussion about SET sound except what we already know. Your argument is valid independently of the type of amplifiers - solid state or tube - as long as we call distortion anything that moves away from strict linearity and have a preference. IMHO the interesting thing is identifying the real causes in detail and correlate with perceived sound quality - and here some people refuse to accept their ignorance of the real reasons.

BTW, you forgot noise!

 

[microstrip]

Ha-Ha-Ha!
Are there any speakers with a flat enough impedance characteristic to have its FR inaudibly changed by a SET amp,given that 0.1 dB change is audible?

Why should we be obsessed with a .1dB meaningless change?

 

[YashN]

Ha-Ha-Ha!

I think we can Ha-Ha-Ha! at your erroneous attempts at self-justification as well, especially if you think you have an informed opinion about Engineering if you aren't an Engineer yourself - no harm done - laughter is the best medicine (from a Reader's Digest recurring column).

Have a good day.

 

[LL21]

Hello LL21


I think because of the higher sensitivity the higher efficiency speaker will tend to be more linear and operating in that range more often than not. As I see it the higher efficiency has benefits all the way down the line as all components will be operating in essentially a low power mode at listening levels we would typically listen at. This is even more important if we listen at anything near life like levels as the 85db speaker is decidedly on the wrong side of the power curve and both the amplifier and individual drivers will be working quite hard at that point. Lets face it a 15db advantage is huge as far as the power levels required for 110dB peaks. Even at an average of 95db you are looking at 10 watts vs milliwatts That's at 1 meter and does not account for distance to the speakers. Distortion levels in higher efficiency drivers can be quite good so an advantage all around.

Attached is a distortion measurement on JBL KS9900 speaker system at 96db you can see that the distortion is under .3% in the midband and quite good over the entire bandwidth.

Rob

Wow...thank you! Very interesting. I note that Chuck, a member here, owns XLFs with the mighty Boulder 3050s...a 94db sensitive speaker with 1500 Class A watts. Talk about effortless and loafing...and possibly, i suppose, on an excellent track towards super low-distortion levels.

I know that i don't 'need' (nor use anywhere near) 160 Class A watts that double all the way down to 0.5ohms (and double again for peaks)...but it i do wonder about what i actually do use if we include every single peak during an orchestral piece played at a 'base level' of 80db-85db at the listening position (ie, peaks in orchestral might hit 115db)

 

[morricab]

I agree that flawed reasoning leads to flawed deductions, but it is your reasoning which has no basis in fact.
Simplicity of circuit has no merit if it results in a non-linear solution, it is poor engineering understanding.
There is evidence, quoted by another poster, that people found the addition of 0.5% 2nd harmonic distortion sounded "more musical" than the undistorted signal. Daft but that is why some people say the same about SETs, the concept of the equipment rather than the musicians being musical is ridiculous.

You are just another person trying to justify your preference to yourself without admitting that it is the added distortion which you like the sound of. IMO that is the only explanation that supports the facts, all the rest is self delusion.

From Geddes (2003):
"This is precisely where the signal-based distortion metrics fail. In our next paper we will show that .01% THD of one type of nonlinear system can be perceived as unacceptable while 10% THD in another example is perceived as inaudible. Even one of these simple examples is sufficient to invalidate THD as a viable metric for discussion of the perception of distortion. Furthermore, 1% THD is not at all the same as 1% IM, but we will show that neither correlates with subjective perception. While some of the signal-based metrics may be “better” than others, it is our opinion they all fall short of what we are seeking."

"One reason that the perception of nonlinear distortion is so complex is that the hearing mechanism itself is not linear and taken as a “system” it is also quite complex. It should thus be expected that it will be a difficult task to ascertain what levels and types of nonlinearity the ear can perceive and even more difficult will be the scaling of the subjective impression of these nonlinear functions."

I suggest you read the rest of the paper: http://www.gedlee.com/downloads/Distortion_AES_I.pdf

and it's followup:

http://www.gedlee.com/downloads/Distortion_AES_II.pdf

0.5% of pure 2nd harmonic added is almost certainly inaudible...

Basically, Geddes findings mirror CHeever's master thesis pretty closely. Both support what I have been telling you and not do support how firmly you feel about the subject.

 

[andromedaaudio]

High eff speaker high power amp , i like it

Its like cruising in a bentley or a ford escort , both do 50 miles an hour but the ride is quite different , but you need the room for it to fully exploit it