Attention AKSA 55, Hugh is thinking in let us see schematics on forum

Status
Not open for further replies.
Jump to Latest
#104
AKSA said:
Tlf, the Earl Geddes work is extremely interesting, I've seen it before. Dave, my thanks for including it in the debate, it absolutely encapsulates the issues.

Here's the reference:
http://www.gedlee.com/distortion_perception.htm

To quote Earl: "While "your results may vary" one should normally find no correlation between perception and the THD and IMD measures. A good correlation with the GedLee metric is evident."
Click to expand...

This work is one of the rare that shows first hints of convergence of perceptual coding with assessment of audio reproduction quality.

There are obviously two approaches - 1) assure that audible distortions *can't* occur, thats what ever lowering THD strives for, and 2) assure that distortions that occur are *not* audible. Well, there is a 3rd one too - create distortions that please the listener.

What Earl really showed is not that THD metric is totally useless. He only offered relatively bad examples of THD and IMD distortions by modern standards to illustrate his point. There is certainly without doubt a threshold of THD and IMD below which distortions are not audible for anyone. Examples he picked are extreme cases to make a wake up call. But if statistical analysis on larger scale would be made, its not at all certain that these examples would show more evidence, but instead would show that approaching THD=0 correlates with inability to distinguish from reference very well.

What his examples show is that in certain circumstances very low THD is overkill, and even very large amounts of distortions can go unnoticed. Well, thats what perceptual coding is all about. Whats worse, he didn't assess audibility of distortions, as would comparison to audible differences with reference show, but he showed different *acceptance* of distortions even if they might cause audible difference. His agenda was that not every THD artifact is heard or Intolerable.

One has to decide clearly, whether what he looks for is reproduction without any perceptible changes to recorded material, or he looks for pleasant representation thats larger than life. If former, then THD approaching zero is certainly a vehicle to the goal, and so is perceptual coding. If latter, then black magic and art is all the way to go.

ps. I analized alittle the wavs. The phan09 (0.1% THD) has bizarre crossover distortion (near every zero crossing adds discontinuity. Analog equipment is hardly even capable of such distortion). That certainly would show normalized THD figures very drastically dependant on frequency of test signal, and also amplitude. The problem could be with single 1khz frequency testing at full amplitude, but if frequency sweep at different amplitudes were used, THD would have shown way higher distortions. The residual exceeds(!) input signal at most silent passages.

The phan03 (12% THD figure) shows softlimiting at highest amplitudes only. That manifests itself only during high amplitude peaks, ie. relatively rarely with dynamic music (but flat in the face with full-range THD test), and is obviously masked by the high amplitude signal. That same transfer function would show way lower THD at low amplitude test signals. Infact, there is 0 residual for amplitudes below -12db - no wonder it sounds okay with real music.

So, I can see clear correlation with THD measurements. Just you can't test at single amplitude (full power) and single test frequency, and you have to normalize. But this is no news to anyone I hope.
 
#105
I have no example of an amp designer, from the least to the most subjectivist, who does not use THD measuremenents when developing an amplifier.

Suppose a designer has a clear idea how to design a new amplifier using only subjectivist recipes which he thinks are good. He builds it and listens to it. I bet he is going to make some THD mesasurements, at one time or another, when finalising it.

I often read THD has no sonic correlation with auditive experiences. Maybe, but everybody uses it at the designing and checking stages.
Is that not contradictory ?

~~~~~~ Forr

§§§
 
#106
Design Criteria

forr said:
I have no example of an amp designer, from the least to the most subjectivist, who does not use THD measuremenents when developing an amplifier.

Suppose a designer has a clear idea how to design a new amplifier using only subjectivist recipes which he thinks are good. He builds it and listens to it. I bet he is going to make some THD mesasurements, at one time or another, when finalising it.

I often read THD has no sonic correlation with auditive experiences. Maybe, but everybody uses it at the designing and checking stages.
Is that not contradictory ?

~~~~~~ Forr

§§§
Click to expand...
Not me I design and listen. If I am satified I finalise my design and make a PCB for it or modify Bongiorno's design, hahahahaha! Any idea forr how much that Audio Precision equipment costs?

😎
 
#107
Elso Kwak

"Not me I design and listen. If I am satified I finalise my design and make a PCB for it or modify Bongiorno's design, hahahahaha! Any idea forr how much that Audio Precision equipment costs?"


I hope you have an oscilloscope to test it.
No need for an Audio Precision. Building a distometer is not that difficult. A good tip, if resolution is insufficient : increase the gain by modifying the NFB resistors.

~~~~~~ Forr

§§§
 
#109
missed mu point

forr said:
Elso Kwak

"Not me I design and listen. If I am satified I finalise my design and make a PCB for it or modify Bongiorno's design, hahahahaha! Any idea forr how much that Audio Precision equipment costs?"


I hope you have an oscilloscope to test it.
No need for an Audio Precision. Building a distometer is not that difficult. A good tip, if resolution is insufficient : increase the gain by modifying the NFB resistors.

~~~~~~ Forr

§§§
Click to expand...


Hi forr, I am afraid you totally missed my point...........🙄 🙄 🙄
 
#110
Re: Design Criteria

wimms said:
This work is one of the rare that shows first hints of convergence of perceptual coding with assessment of audio reproduction quality.
Click to expand...

thanks, wimms, for an excellent analysis.

Elso Kwak said:
Not me I design and listen.
Click to expand...

there may be a distinction between people who design commercially and who design for pleasure (which I assume you are. If I am wrong, I apologize).

When we design for ourselves, we know precisely what the "customer" needs and it isn't hard to test the amp in the "market" and know exactly how it is received.

When we design for others who pay for our products, we don't have that luxury. And we also know that any failure in designing a product that is to be well received by the customers is going to be felt by the designers adversely. So we will need to "approximate" what the market / customers like, or what "good sound" means.

I venture that designing for THD is a compromise between a workable solution and our desire to produce the best sounding amp for the market.
 
#111
forr said:
Suppose a designer has a clear idea how to design a new amplifier using only subjectivist recipes which he thinks are good. He builds it and listens to it. I bet he is going to make some THD mesasurements, at one time or another, when finalising it.
Click to expand...

Yes he is. But a competent designer isn't going to pay much attention to the single number THD. He is going to look at the THD spectrum chart(s) that the single number comes from.

As far as THD number approaching 0, this is where you will find some of the very worst amps. To get these numbers large amounts of feedback are most often applied with the result that the vanishing low number is made up of higher order harmonics which are detectable at much, much lower levels (ie an amp with 1% 2nd harmonic is likely to sound WAY better than an amp with 0.1% 6th or 7th order harmonics)

(how many times have you seen Nelson say something like, "i got the THD down to 0.001, but the version with 0.1% sounded better)

dave
 
#112
PLANET10
"i got the THD down to 0.001, but the version with 0.1% sounded better"

How many times have we heard that ?
Does it mean the 0.001 THD amp is bad ?
Or does it mean that the preference is for a higher level of harmonics, because it makes the sound richer ?
Who has made blind comparisons with a "Self's niceness box" and a so called "good sounding amp" ? I would be interested to know some references for such experiences.

Low THD amps are usually rather similar to op-amps and most recorded material has gone through many op-amps. Do many people avoid op-amps at the recording stage ?

Take thirty identical 0.001% amps. Connect them in series with an attenuator equal to the gain between each . Listen to the result.
Do the same with thirty 0.1% amps. I bet the output of the 0.1% series will be not so nice compared to the 0.001% series.


I just see so many misconceptions about high frequency harmonics associated with high feedback that I won't discuss it anymore. This has been studied many times. And once again, if it were true, any material having passed through many op-amps would have a huge level of high frequency harmonics. It has not.

~~~~~~ Forr

§§§
 
#113
forr said:
Or does it mean that the preference is for a higher level of harmonics, because it makes the sound richer ?
Click to expand...

It is typically 2nd & 3rd that make an amp sound richer. Higher harmonics are grating to the ear.

Here is an example of a spectrum that collapses to a low number, but has a scary amount of high order harmonics.

thd.png


dave
 
#114
planet10 said:

As far as THD number approaching 0, this is where you will find some of the very worst amps. To get these numbers large amounts of feedback are most often applied with the result that the vanishing low number is made up of higher order harmonics which are detectable at much, much lower levels (ie an amp with 1% 2nd harmonic is likely to sound WAY better than an amp with 0.1% 6th or 7th order harmonics)
Click to expand...

Partly i agree, but you dont necesairly have to increase open loop gain to reduce distortion ( witch doesnt solve it at all, just replaces the problem ), other ways off solving these issues have been around for about 2 decades. But only very occasionally used.

And therefore one cannot simply say that amps off 1% THD sound better then 0.001% before knowing how they reached that result.

That said single tone measuments don't tell you all there has to be known about an amp to predikt how its sounds.

Rudy
 
#115
Rudy said:
And therefore one cannot simply say that amps off 1% THD sound better then 0.001% before knowing how they reached that result.

That said single tone measuments don't tell you all there has to be known about an amp to predikt how its sounds.
Click to expand...

exactly. The single numbers just don't have enuff context to be meaningful.

Now if you had the spectrum of the 1% & the 0.001% amp (at a number of frequencies) you would have a much better chance...

dave
 
#117
planet10 said:
As far as THD number approaching 0, this is where you will find some of the very worst amps.
Click to expand...

that's why the argument gets illogical: just because there are terrible sounding amps with close to 0% thd doesn't by itself mean that thd isn't a meaningful indicator of performance. ASKA and I discussed that just a few posts ago.

You have not presented any facts or valid argument that thd isn't correlated to performance (and the "evidence" thrown out by Gedlee was fairly analysed by wimms to which you had zero counter argument).

Yet, you keep repeating "i got the THD down to 0.001, but the version with 0.1% sounded better". Don't you realize that no matter how many times you repeat it, it isn't factual unless you can back it up with facts?
 
#118
tlf9999 said:
Yet, you keep repeating "i got the THD down to 0.001, but the version with 0.1% sounded better".
Click to expand...

... not something i said, just an illustration from a very respected amp designer. I ignore THD numbers (i'll happily look at the charts). I'm more concerned with whether i enjoy listening to an amp.

I'm not going to change your mind obviously -- i'll just go back to enjoying my amplifiers -- the most enjoyable of which no doubt has the worst measured single number THD (and the worst, the best -- i know there isn't a negative correlaltion either -- one of my quite good amps should measure pretty well).

My experience has told me that single number THD is meaningless, and maybe others will benefit. We are starting to see solid scientific research to support this -- i do hope Hugh can get his research project off the ground and 10 years from now it will "be obvious".

Don't you realize that no matter how many times you repeat it, it isn't factual unless you can back it up with facts?
Click to expand...

I do... just as no matter how many times you repeat that THD (single number) is a valid metric does not make it factual.

dave
 
#120
planet10 said:


... not something i said, just an illustration from a very respected amp designer. I ignore THD numbers (i'll happily look at the charts). I'm more concerned with whether i enjoy listening to an amp.

I'm not going to change your mind obviously -- i'll just go back to enjoying my amplifiers -- the most enjoyable of which no doubt has the worst measured single number THD (and the worst, the best -- i know there isn't a negative correlaltion either -- one of my quite good amps should measure pretty well).

My experience has told me that single number THD is meaningless, and maybe others will benefit. We are starting to see solid scientific research to support this -- i do hope Hugh can get his research project off the ground and 10 years from now it will "be obvious".



I do... just as no matter how many times you repeat that THD (single number) is a valid metric does not make it factual.

dave
Click to expand...


Planet 10,

I agree with you!
BTW I am pretty sure I have the same respectfull amp designer in mind as Planet 10 before he mentioned it in his recent post as this commersial designer I have in my mind have made a simple and clear statement in the matter easy for everyone to grasp just for some year ago right here on DYIaudio if my memmory serves me right I would say.

The kind of arguing I have seen started after around post #60 which has escaletd in only two days into another close to 60 posts is depressing.

TLF9999, to be frank, you have been around here since May, may I suggest you to prove and/or provide something yourself instead of throwing in the face in every second or so post from you something like "can you prove it", and that even to some of the nicest gents on this forum.

It's quite obvious from your postings that you don't understand what the THD spectrumcomposition means to our ears psychoacoustically.
Distortion discussions have been treated quite heavily before(you registered on in May) here on DIYaudio, so may I suggest you to use the button instead of making heavy postings with lot of "empty" arguings which doesn't contribute anything more than flaming up the thread going too much OT which is already way from the subject "Attention AKSA 55, Hugh is thinking in let us see schematics on forum".

Cheers Michael
 
Status
Not open for further replies.
Top Bottom