Sound Quality Vs. Measurements

Status
Not open for further replies.
Is there a difference between reproduction accuracy and reproduction quality?

I'd really enjoy seeing a schematic and BOM for an AP analyzer. Good test equipment should take what's there and send it to a display; good audio equipment should take what's there and send it to a speaker.:2c:
 
Hi,

Good test equipment should take what's there and send it to a display;

Wrong. Any test equipment has the job to give a precise (precision of determines cost) answer or set of quantive answers to a set of predefined qualitative questions (e.g. how much harmonic distortion).

If it answers the question accuratly it is good test equipment. If the question is wrong (as in the quality being quantified is in fact meaningless in the context in which it is employed ) it is still good test equipment, even though it is in effect an expensive boat-anchor for this particular subject. I would argue that the AP2 qualifies mostly as such an expensive boat-anchor for Audio, even though I regularly get to use one (it answers mostly the wrong questions)...

good audio equipment should take what's there and send it to a speaker.:2c:

Why? Given what the speaker is going to do with it (distortion, compression, frequency response distortions etc. et al)? Where the point of this?

It is like the Pharisee's in the parable that Yeshvha guy told, of them straining at the gnat in their wine (that is publicly straining their wine so they would not unintentionally violate jewish dietary law; used as an illustration how the Pharisee's would make a great show of obeying small laws in public) while gulping down the camel (that is to violate grossly the same laws when out of sight).

Similarly we see here a discrimination against comparably small flaws in measured performance in the sense of traditional measurements in electronics being painted as unacceptable, while the much larger flaws of the same nature commissioned by the speaker are simply swept under the rug. Where is the sense in that?

I would suggest that good audio equipment (including speakers) should provide a realistic reproduction of a musical event and allow an easy "suspension of disbelief". This so far has not been shown to be reliably accomplished by equipment that exceeds in certain traditional measurements over other equipment.

In fact, often the reverse seems to be the case.

So either there hidden variables not accounted for by traditional measurements and views, or people just like "bad sound".

I find this latter position both unconvincing and tautological, because if people like "bad sound" then by definition this "bad sound" would be actually "good sound", as it simply means people disliking what technocrats want to push on them. So it in fact the "people like bad sound" explanation in reality underlines the "hidden variable" one, one might even successfully argue that the two are the same.

Ciao T
 
Wrong. Any test equipment has the job to give a precise (precision of determines cost) answer or set of quantive answers to a set of predefined qualitative questions (e.g. how much harmonic distortion).
I propose that the above is wrong. Show me a piece of test equipment with a "question" input. It is the operator of the equipment that performs the question and answer stuff. Again, the equipment should just show what's there. To be interpreted.
The rest of your response just means to me that you see a difference between the accuracy and quality that I began my post with.
For anyone who shares that opinion (and I really have nothing against that view tbh), then arguing against measurements (and vice versa) is futile and irrelevant.
BTW you're making a straw man of "bad sound." If quality isn't accuracy, then non-accuracy is not therefore bad sound. It could very well be "quality sound."
As for hidden and not accounted for "variables," I did not discount them, neither expressly nor implied. Progress, not perfection, is the only reasonable goal.
 
Hi,

I propose that the above is wrong. Show me a piece of test equipment with a "question" input.

AP2 (semi-joking).

The question input is the actually quantety being measured, that is a Volt-Meter measures voltages, if you want to (for example) measure temperatures you first need to convert them into voltages to measure them using a volt meter.

So the questions are designed into the test gear. You cannot for example use a traditional THD & N Meter to measure TIM.

With the AP2 and many modern test instruments you instead have some hardware and some DSP which can be used for many different measurements, however there is still the need to define a question, as in "how much THD".

It is the operator of the equipment that performs the question and answer stuff.

Not really. Let us take my previous example of the classic needle display THD&N Meter (I still have one around incidentally). The question is designed, the answer is read off.

If I want to have a different question answered (e.g. how much TIM) I cannot use this THD&N Meter, because it does not answer the question.

The key here is that test equipment can only answer pre-defined questions, generally it cannot answer questions about other quantities than these it is designed to measure.

Again, the equipment should just show what's there. To be interpreted.

There is no such test equipment for audio I know. In all cases the question and interpretation is designed in. For example, there are some forms of distortion (I call them fuzzy distortion) that are clearly visible on an analogue 'scope with the naked eye (that is they exist in reality), yet you cannot use an AP2 to measure them...

BTW you're making a straw man of "bad sound." If quality isn't accuracy, then non-accuracy is not therefore bad sound. It could very well be "quality sound."

That was the point I was trying to get across. If an amplifier with low measured THD sounds bad according to listeners it sounds bad. If one with high measured THD sounds good according to listeners it is good. Yet our technocratic elite here at DIYA will insist that the second item is "bad" and the first is "good".

As for hidden and not accounted for "variables," I did not discount them, neither expressly nor implied. Progress, not perfection, is the only reasonable goal.

This I agree with, progress is desired.

As for the hidden (and so far unaccounted and therefore not measured) variable, we do find that often an amplifier with greater distortion than another is perceived as better, yet adding deliberate distortion to a "clean" amplifier usually fails to improve it's sound (try it).

This suggest that in fact the difference is not in measured distortion.

Yet people keep going on futile over low measured THD when in fact this has very little correlation with "good sound" and when most speakers have hugely gross distortion, compared even to the often maligned SE Tube Amplifier...

Ciao T
 
For example, there are some forms of distortion (I call them fuzzy distortion) that are clearly visible on an analogue 'scope with the naked eye (that is they exist in reality), yet you cannot use an AP2 to measure them...

Perhaps you might not be able to, but most competent engineers could. If you can see it on a scope, it will cause a difference in the frequency spectrum.
 
The question input is the actually quantety being measured, that is a Volt-Meter measures voltages, if you want to (for example) measure temperatures you first need to convert them into voltages to measure them using a volt meter.
OK I have a better understanding of your analogy.
The key here is that test equipment can only answer pre-defined questions, generally it cannot answer questions about other quantities than these it is designed to measure.
Key, yes. Definitely.
There is no such test equipment for audio I know. In all cases the question and interpretation is designed in. For example, there are some forms of distortion (I call them fuzzy distortion) that are clearly visible on an analogue 'scope with the naked eye (that is they exist in reality), yet you cannot use an AP2 to measure them...
A concrete example of your previous statement.

The next bits directly relate to my accuracy and quality question. If reproduction accuracy is the goal, then high THD/TIM is bad.
There are countless combinations of source-amp-loudspeaker and so a thousand anecdotal opinions don't mean much to me. Testing provides the meaningful data. And every test includes a measurement AFAIK.
My own analogy, though, would be something like... some say to-may-to and some say to-mah-to, and they're using that as a basis to argue over how to make the sauce.
 
Sy,

Perhaps you might not be able to, but most competent engineers could. If you can see it on a scope, it will cause a difference in the frequency spectrum.

Rest assured that you cannot directly measure it using an AP2 in either analog or digtal mode. It is not capable to be set up to give me a measurement where I can say "X" or "Y" amount of FD.

Yes, you can get indications of the presence of FD and it's audio band effects, but only indirectly and with effort (e.g. you need to know it is there, look actively for it and devise a test regime to do so, but even then you cannot be sure what caused the measured effect), but that is not what I call measuring something.

Ciao T
 
Yet seeing and interpreting those differences is what a competent engineer would call "measuring something." You certainly can call it "peanut butter and jelly" if you like, but that won't really accomplish anything. If there's a difference in the time domain, there is a difference in the frequency domain. Which one uses is a pretty trivial choice, which is why any engineer has both a scope and a spectrum analyzer.
 
Hi,

If reproduction accuracy is the goal, then high THD/TIM is bad.

The first problem with your statement is the amorphous, undefined term "high". What is "high THD/TIM". What is our yardstick?

A number set on a green table?

A number derived from the human hearing to represent audibility? If so one based on the average or the worst case?

The fact is that "zero distortion" is not possible. So "low" and "high" must be seen with a point of reference.

And especially THD is an incredibly poor indicator.

Earl Geddes once posted four tracks of music. One had been subjected to a process that produced 0.1% THD, the second to one that produced 9.6% THD and the third produced 12.6%.

As it so happened, even on extremely low end gear (PC Speakers) the 0.1% THD track was instantly perceived as extremely distorted, yet even on fairly high end gear the 9.6% THD track was essentially impossible to tell from the undistorted reference with any reliability, the 12.6% THD track was identifiable as being a little too warm sounding.

So, in one case 0.1% THD is too much and in another 9.6% is inaudible, which tells us what precisely?

My own analogy, though, would be something like... some say to-may-to and some say to-mah-to, and they're using that as a basis to argue over how to make the sauce.

Please understand that measurements provide data, not information. Information is data placed in a context.

If the context is wrong, the resulting information is wrong. Again, in my example the context is "question".

If my question is "does this device have 1) audible distortions and 2) objectionable distortions" then THD is not useful. And I would suggest that when it comes to distortion my question is the correct one and the one answered by THD&N is the wrong one.

Ciao T
 
Well Thorsten, you fight the 'good' fight. I am not able to, unfortunately. I am NOT for deliberately added distortion, but 2'nd harmonic has been shown to be tolerable to quite high levels, throughout the history of audio reproduction. Even 3'rd harmonic is reasonably tolerable until you get above 3% or so, maybe even more on peaks. This was proven for about 50 years with analog master tape recording that produces 3'd harmonic almost exclusively. It always amazed us as to how good we could make solid state electronics, YET the loudspeakers and the sources, typically analog tape or vinyl records, had so much higher distortion. How could making 'better' electronics make any difference at all? Because we could still hear the differences. This is why, Bascom King, another engineer, and I went to see Richard Heyser at his home, about 40 years ago. Richard told us that it was 'negative feedback' OMG! That made the difference. He said that he proved it more than a decade earlier with a power amp design that he made for JPL, allegedly to go to the Moon. There is an AES preprint of this amp design in the AES archives. Talk about a crazy design. We should make it up and try it, ourselves. Looks like something like Nelson Pass might make as a joke. He said that he just tried it on his K-horn for the heck of it, and found that it sounded better than anything else he had around. HE pinned it down to 'negative feedback', I didn't. I just know from experience that he was on the right track.
For the record, Bascom King designed the major power amp that recently got ' Power Amplifier of the year' in TAS, which is based on a unique design of his that he discussed with me, some 40 years ago.
He, Peter Madnick and I (by designing the JC-80 that it was based on), produced the preamp that got the 'Preamp of the year' award in the same issue of TAS, Jan. 2012
 
The fact is that "zero distortion" is not possible. So "low" and "high" must be seen with a point of reference.
Not physically possible. That doesn't prevent zero from being a reference.
Earl Geddes once posted four tracks of music. One had been subjected to a process that produced 0.1% THD, the second to one that produced 9.6% THD and the third produced 12.6%.
Mr. Geddes has probably forgotten more about electroacoustics than I'll ever know. I can only search for that thread and meagerly attempt to figure out what was happening.
If your question is "does this device have 1) audible distortions and 2) objectionable distortions" I would say if reproduction accuracy is the goal, the context, if you will, then audible distortions, whether objectionable or not, are reproduction inaccuracies. Calling (or perceiving) a distortion pleasant/spacious/warm/alive/whatever doesn't make it any less inaccurate.
 
He has as much right to state his view as those in this forum who, unequivocally, state that a form of interconnect can alter the behaviour of individual electrons flowing through it.

Yes, this is sarcasm and yes, I am amused by the new branches of physics which are, seemingly, discovered, almost daily, in these forums.

I also apologise to all of those mathematicians, scientists and engineers who toil daily in the misguided belief that they are actually contributing something to human knowledge when it is quite clear that all you need is self-belief and no knowledge whatsoever.

Frank
 
ThorstenL said:
If an amplifier with low measured THD sounds bad according to listeners it sounds bad. If one with high measured THD sounds good according to listeners it is good. Yet our technocratic elite here at DIYA will insist that the second item is "bad" and the first is "good".
This argument technique is called "attacking a straw man".

So, in one case 0.1% THD is too much and in another 9.6% is inaudible, which tells us what precisely?
We should design our equipment to give precisely 9.6% THD at maximum output? (or normal listening level?)
 
A half-wave pulse has infinite bandwidth (although finite slew rate), so all (finite-bandwidth) amps will modify it in some way. Testing audio amps with a non-audio test signal is of course an interesting exercise, but you then have to try to read across to how the amp will behave with an audio signal.

"frequency compression in the initial wavefront": this is mixing frequency and time domain descriptions. Could you show us some of your results? My guess is that most of what you are seeing is simply low-pass filtering, but the phase response around the roll-off point could have some effect. The really interesting part would be any residual non-linear effects, as you surmise.

Quite right, but I am looking for SOMETHING I can point to that is a repeatable test to show there is a difference in the transfer function. I don't have a way to capture this in a picture yet as I was using an analog scope. I only described what I saw, not what seemed to cause it. I will get back to this after I play with the Hexfred diodes I ordered and "fresh" caps for some of my 30 year old amps, and oh yea, I am in the middle of another set of speakers to optimize. This would be easy if I did not have to work for a living too.

With all due respect to the current amp designers, why do my old Rotel RA-840 and RB-941's pass my wife's critical ear and a lot of it's contemporaries do not? Failing are amps from Adcom, Parasound, Nad, Hafler, Aimor, and B&K. My Creek seems OK. None of these were considered slouches! If I can remember back that far, she passed the Aaragon and a small Krell but they were out of my price range. I would love to try a Naim, Linn, Cambridge, Music Fidelity,or a Bryston but alas, not even close for used one but I keep bidding. :D
 
With all due respect to the current amp designers, why do my old Rotel RA-840 and RB-941's pass my wife's critical ear and a lot of it's contemporaries do not?

It may have nothing to do with the amps themselves. What you want to do (if you're really interested in nailing it down) is to have her do a rigorously level-matched comparison (that means within 0.1dB), at levels under clipping, without YOU (as well as her) knowing which amp is switched in when. Double blind is particularly critical with married people- see my favorite anecdote in the last issue of Linear Audio.

If she can distinguish them then, you have a valuable asset on your hands which you can probably monetize.
 
Status
Not open for further replies.