Sound Quality Vs. Measurements

Status
Not open for further replies.
Including our ears, as for example anyone playing some of the Telemann recorder duets can discover (it's been suspected that Georg Phillip selected pitches sometime to intentionally produce a difference-frequency bass line).
You might find that paper at least interesting if not exiting : http://igitur-archive.library.uu.nl/med/2011-0204-200555/Book Heerens de Ru EN.pdf
EDIT: For a shortcut, start at pg. 25


There is a piece of software to recreate the experiments from the paper, available from the author's web page :
three compartment cochlear model

The SW is also very good to test audibility of phase distortion and absolute phase, just detune the harmonics for a continuous shift, or use the phase offsets. In my limited understanding the text doesn't offer a solution to those effects, but maybe I'm just missing something...
 
Last edited:
diyAudio Member RIP
Joined 2005
You might find that paper at least interesting if not exiting : http://igitur-archive.library.uu.nl/med/2011-0204-200555/Book Heerens de Ru EN.pdf
EDIT: For a shortcut, start at pg. 25


There is a piece of software to recreate the experiments from the paper, available from the author's web page :
three compartment cochlear model

The SW is also very good to test audibility of phase distortion and absolute phase, just detune the harmonics for a continuous shift, or use the phase offsets. In my limited understanding the text doesn't offer a solution to those effects, but maybe I'm just missing something...

Sehr interessant! Thanks for that. Did they ever get published in a journal, finally?
 
news to quite a number of people...
all it takes is to clearly measure is a cheap condenser mic and a classic bookself speaker with long throw 4-6" mid-woofer

Please do show me intermodulations between 50Hz and 2000Hz in a two-way speaker and compare them with what an amp can produce.
Or even closer frequencies if you want... to go on a single speaker.
Sure, there are plenty of THD (mostly lower orders), due mostly to the mechanic friction, but you probably you assume they are the same thing like IMD and... you miss completelly my point.
 
Last edited:
Hi,

Please do show me intermodulations between 50Hz and 2000Hz in a two-way speaker and compare them with what an amp can produce.

First, please review the poster from Klippel I have linked. It shows how to measure the intermodulation and how to predict it from the drivers behaviour.

Sure, there are plenty of THD, but you probably you assume they are the same thing and you miss completelly my point.

In a simple non-linear system IMD and HD are results of precisely the same non-linearity and have the same levels. So if we have at 1KHz -60dB H2 and -80dB H3 and present the DUT with equal amounts of 1KHz and 1.133KHz giving the same peak level as the HD measurements the 2nd order difference tone will be at 133Hz with -60dB and and the 3rd order one at 876Hz and 1266Hz with -80dB.

Come on, this really basic stuff and essentially geometric/mechanical in nature. It will hold true for any non-feedback system and to a certain degree also for ones with correctly inverse feedback.

Ciao T
 
ThorstenL said:
In a simple non-linear system IMD and HD are results of precisely the same non-linearity and have the same levels.
Yes! This simple fact seems to have escaped far too many people.

You can deliberately engineer a system with IMD and HD at different levels, and there are frequency-dependent mechanisms which can affect them differently, but as Thorsten says the straightforward situation is that they are generated in the same way.
 
Sure but their 5ths and 7ths wind up some place i can't probably hear anyway. They sure can sometimes sound nasty, no doubt.:eek:

Gotta admit i was thinking of harmonics of low frequencies. If they come from upstream in the electronics, then you've got problems because of the energy component - more than 50% of energy in music is below 1000 Hz - because they show up in 700 Hz - 6kHz range. And we're really sensitive to sound from @2kHz to @ 5 -7 kHz.....

That's why I think you can hear surprisingly low amounts of distortion in an amplifier from a speaker that doesn't seem, by comparison to the amp, to have very good distortion figures.

On the surface, but what about IM? My working theory on my wife's uber hearing is high order distortions causing aggravation of tweeter issues above hearing, but causing IM that is down in the critical 4K range. There is SOMETHING going on.
 
diyAudio Member RIP
Joined 2005
Working through Self's second book, Small Signal. Just one view, but he is pretty hard core: It's the circuit, stupid. His statements on "if you can't measure it it, you can't heard it" still leave us with which measurements are the ones that matter.

I detect a mellowing in Doug's style and this is a welcome change. He's done a lot of work over the years and all of it that I know of has been highly competent, some of it quite inspired (I recall an auto-bias circuit for power amps that was very ingenious; I think it was in Electronics World).

He does like to write, obviously. About the only mild criticism I would offer is that he presents a lot of discussions and circuits without all that much history and attribution. Of course this is a lot of work, and it's quite a lot of work to write a book as it is. Nonetheless, there are sections here and there where I look to see when he's going to mention an older publication, and it doesn't appear.
 
Hi,

Working through Self's second book, Small Signal. Just one view, but he is pretty hard core: It's the circuit, stupid. His statements on "if you can't measure it it, you can't heard it" still leave us with which measurements are the ones that matter.

While in terms of a philosophical principle, DS is kind of right, he got things backwards.

It should be "If you have heard something it is possible to devise a measurement to quantify the quality that causes what was heard.".

If DS rendered his position as he wrote it, it would imply that the absence of a suitable measurement makes an effect inaudible, the basic logic fallacy of such a position is sufficiently obvious that should not need illumination.

However, what bis perhaps worse, based on what I have read from DS over the decades, including LTE's and Answers to LTE's in Wireless World, his books as well as his contributions on line, the man is actually really convinced that if our rather limited battery of tests available and common at this point do not show it it cannot be heard. Again, the failure in basic logic of such a position needs no further illumination.

Incidentally, what I find DS's work most useful for (up to a point only unfortunately) is the hard data on many phenomena in amplifier design that he has collected over the decades. It saves time doing the tests oneself.

That said, I still find Bob Cordell's book more useful as he often includes not THD, but spectrae or at least indications of harmonic distribution, which DS almost never does, despite having the instrumentation to do so.

Plus I would again IN THE STRONGEST POSSIBLE TERMS suggest reading Samuel Groner's commentary on Self's Amplifier book.

Ciao T
 
Scott,

...Brits..., the audio community there seems a little insular.

Well, they DO call it the British Isles, don't they... :D

That said, lack of attribution of older works or even works of which ones own are derivative seems as common in the former British colonies in North America (no flames please) as in Britain itself...

Ciao T
 
Be careful, bcarso. Don't think that hard work is exactly the equivalent to sonic successfulness. I don't write books, but I seem to be pretty successful in that department, mostly because I trust my ears and those of my colleagues, over some 'measurement'.

Exactly what I've been saying for two decades now.

Without wishing to belittle any part of the process, in the end, it all comes down to the designer's own hearing. He does the final voicing according to his own hearning and taste. He is the one who says: "Fine, no more, this is IT!".

And he would be wise, as John did, to include the hearing of his associates, and/or people with hearing he trusts; it's ALWAYS good to have several opinions, because much as I trust my own ears, I do not think of my own hearing as absolute.

I read in an interview once that Dan d'Agostino does that by using a pair of headphones especially tweaked for his own hearing, because he says he's most interested in noise and grime at low levels, because that's where most of the trouble is in his view, and no loudspeaker can help there like a pair of headphones can.

And the devil can indeed be in the details. Two months ago, I refreshed (exchanged all the electrolytics) on my old H/K 6550 from 1993. H/K has the nasty habit of using only pots and trimmers they can guarantee with their lives are the worst junk you can find on the makret - after 6 or 7 of their units, I think I have the right to say that. So, along the way, I decided to change their volume pot, from a cheap'n'nasty to an Alps Blue. Since then to this day, I wonder at the difference, finding it hard to believe that changing just one pot can bring about such a radical difference in sound - from a precise but fairly flat sound to all the ambience I could reasonably hope from such a device, at such a cost.

I take this as an object lesson - you can make a great amp, very linear, with low overall feedback, the works, but fail at just one critical point, and you're done, it's all gone to waste.

An accountant (bean counter) is not likely to understand the price difference, for him, a pot is a pot, period. This is where the designer should have the last word, but all too often doesn't.
 
Last edited:
Sehr interessant! Thanks for that. Did they ever get published in a journal, finally?
AFAIK, not yet. I really hope they make it, as soon as other researchers will have taken efforts to back up their hypothesis (or disconfirm it, worst case).

-------:------

Looks like they get ignored here, too :-/

Maybe I should quote the very first sentence of chapter 8, Conclusions (emphasis mine):
Heerens & de Ru said:
Our careful analysis of the results that were obtained in the auditory pathway experiments that were executed by Wever and Lawrence [16], leads us to the conclusion that the auditory sense differentiates and squares the incoming sound pressure signal.

Personally this statement struck me like a flash of lightning, with regard to its consequences....
 
Last edited:
I'll retread a bunch here. If two amplifiers sound different, at LEAST one of them is wrong. Knowing there are understood issues with every part in an amp, the both are wrong is probably correct. Now, what is it that is wrong on one amp vs the other that bothers each of us more or less? Is this a masking issue, a balance issue? Physical differences in out hearing or training?

Why can't we do this, or won't we do this, with a "simple" amplifier? Is it that the input and output loading is such a larger variable that it is moot? Are the differences too hard to explain to one who is not an experienced engineer? Are all the deficiencies so great that it is a matter of taste which are reduced at the expense of others? Kind of like JFET vs BJT for inputs. Both have advantages and both have deficiencies that can be described in technical terms. Both can be executed to levels of common measurement that suggest we should not be able to tell the difference.

Now, back to understanding why LTSpice FFT shows about the same distortion on input as output.
 
I'll retread a bunch here. If two amplifiers sound different, at LEAST one of them is wrong. Knowing there are understood issues with every part in an amp, the both are wrong is probably correct. Now, what is it that is wrong on one amp vs the other that bothers each of us more or less? Is this a masking issue, a balance issue? Physical differences in out hearing or training?

Why can't we do this, or won't we do this, with a "simple" amplifier? Is it that the input and output loading is such a larger variable that it is moot? Are the differences too hard to explain to one who is not an experienced engineer? Are all the deficiencies so great that it is a matter of taste which are reduced at the expense of others? Kind of like JFET vs BJT for inputs. Both have advantages and both have deficiencies that can be described in technical terms. Both can be executed to levels of common measurement that suggest we should not be able to tell the difference.

Now, back to understanding why LTSpice FFT shows about the same distortion on input as output.

Why must both be wrong? Why can't one be more or less right, and just one off?

In this scenario, you face one more problem - which one is right, and which one is off?

So far, it appears we can't rely on classic distortion measurements because the one with higher THD levels may well sound better nevertheless, assuming that the THD levels on both are below say 0.07% PEAK.
 
Hi,

I'll retread a bunch here. If two amplifiers sound different, at LEAST one of them is wrong.

Actually, even if they sound the same, at least one is wrong...

That is to say all amplifiers alter the signal, so they are wrong, electronically recording music alters the "signal".

With respect, I submit that this search for an Amplifier (or any other part of the recording and playback chain) that is "Not Wrong" is futile and leads into an impasse where you paint yourself into a corner.

The key is make that despite being "wrong" provides a credible illusion.

In this way "serious" audio and music recording are more alike to Magic (that is Magic with a "c", as in creating a credible illusion) than what most would call science (even though I'd call it science, but then I call alchemy and magik sciences).

Now, what is it that is wrong on one amp vs the other that bothers each of us more or less? Is this a masking issue, a balance issue? Physical differences in out hearing or training?

I suspect all of the above plus some not on the list.

Why can't we do this, or won't we do this, with a "simple" amplifier? Is it that the input and output loading is such a larger variable that it is moot? Are the differences too hard to explain to one who is not an experienced engineer?

This is quite a minefield, watch your step. Seems someone even planted a bunch of claymores. Better put that body armour and helmet on and pray real hard. I suspect there are snipers hiding too...

And no, I'm not kidding.

For example John Curl, Charles Hansen (as well as others) have been suggesting for ages that in their view large amounts of negative feedback have a negative effect on sound quality. Those who make these suggestions have a lot of experience, many award winning audio designs to their credit and have been in business for quite a while, so they must do SOMETHING right.

Are the theories put forward as to the why of these problems the last word? No, they are attempts to understand and work from theoretical foundations and to somehow reconcile what we believe we know about electronics. I readily admit that I am completely ignorant as to how electronics really work, seeing that we all cling to that idiotic notion of electron particles that do this and that and the other, but I do know that the models we use and learned in our formal education are often simplified to the breaking point.

As a result this group often comes across almost like mystics.

On the other hand, the ABX Mafia has been spending decades to prove that there are no audible differences that cannot be put at the door of distortion and then there are those who (like Mr. Self) suggest that what they cannot measure cannot possibly be heard.

They basically have already pre-judged the state affairs using very limited (if any) empirical work (their little ABX Test scam has nothing to do with empirical work, outcomes are pre-determined anyway) and merely work on trying to convince everyone that their theories (regardless of experimental proof) are right. Nevertheless, this crowd often comes across as quite a bunch of dry rationalists.

Finally, we have a lot of good science on a range of audible factors (e.g. HD/IMD Audibility) that is studiously and steadfastly ignored much across the board, as the mystics know of it but also know it does not go far enough while the self proclaimed rationalists cannot accept it as it would totally spoil their little theories and scams.

And trust me, the last place you want to be is in this minefield in the no-mans land between the deeply dug trenches of the "objectivists" and "subjectivists". Very uncomfortable and lonely there.

Are all the deficiencies so great that it is a matter of taste which are reduced at the expense of others?

To a degree, I suspect yes. I grew up with Tube Radios fitted with Full Range Speakers, analogue broadcast sources, LP and magnetic tape and much un-amplified acoustic music. So I suspect my hearing adapted and I am by far more tolerant to the faults of such devices, if indeed these faults are audible.

On the other hand I never developed this tolerance for the faults of digital recording, solid state amplification, multiway speakers etc. So to me the sonic faults of these devices are simply unacceptable.

Kind of like JFET vs BJT for inputs. Both have advantages and both have deficiencies that can be described in technical terms. Both can be executed to levels of common measurement that suggest we should not be able to tell the difference.

No, I disagree, there are several areas that suggest that there are several specific areas where J-Fets have dramatic advantages, be it in the presence of moderate levels of RF (which J-Fets will not demodulate to audio, while BJT's will and do so) and if we have appreciable source impedance (e.g. passive preamp, integrated Amp with volume control etc.) J-Fets are by far more tolerant of source impedance than BJT's.

This may not show in traditional bench tests, as it is rarely considered but the effects are measurable and quite material, one merely must take the time and effort to measure them...

Now, back to understanding why LTSpice FFT shows about the same distortion on input as output.

Maybe you made the perfect Amp?

I noticed many simulators do not have very good "oscillators" build in. I believe LT-Spice allows you to use a wave file as input (can it also output a wave file?) so nothing stops you creating a 24Bit / 384KHz Sample rate 1KHz sinewave in some sound editor software...

Ciao T
 
Status
Not open for further replies.