Objectivist or Subjectivist? Give Me a Break

[jkeny]

Ah well, that's where you find yourself when you don't trust your ears - in the land of flawed measurements with no answers!
Why don't you try living with uncertainty - you do it all the time in the rest of your life?

 

[Groucho]

Ah, right so dynamic test signals might trigger issues & reveal more than periodic ones, right? What test signals are used in the measurements you put so much faith in?

I was allowing you the small possibility that a DAC could measure well in certain tests due to good fortune, yet badly at a different temperature, say, due to poor implementation. I am trying to pre-empt the nit picking but failing! If you have some examples of where measurements fail fundamentally could you volunteer that information?

 

[Groucho]

Ah well, that's where you find yourself when you don't trust your ears - in the land of flawed measurements with no answers!
Why don't you try living with uncertainty - you do it all the time in the rest of your life?

This discussion started with my saying that I am reasonably certain that I don't have to worry about DACs. And no, I don't trust my ears in sighted tests on issues that are 'subtle'. I can see how if I did trust them it could be a road to heartache and ruin, always persuading myself that I needed to do one more tweak, churn one more piece of equipment.

What are the flaws in the tests? Do tell.

 

[jkeny]

I've already given you a real-world example which you reject.

Your logic really is flawed - you will accept that a battery of tests are needed to characterise an amplifier's transfer function but reduce this to only two for a DAC's transfer function. Can you please explain your logic?

 

[Phelonious Ponk]

Tim,
You already asked it and I have already answered. No need to repeat.

Sorry, I must have missed it. I remember a recant of the exaggeration "thousands," but I don't remember seeing any links or reports from even "tens."

Tim

 

[jkeny]

.... I could conceive of a situation where a DAC performed well in tests, but certain signals caused instability due to inadequate power supply bypassing of the output buffer or something, and the test didn't happen to trigger the problem. .......

I was allowing you the small possibility that a DAC could measure well in certain tests due to good fortune, yet badly at a different temperature, say, due to poor implementation. I am trying to pre-empt the nit picking but failing! If you have some examples of where measurements fail fundamentally could you volunteer that information?

Your second quoted post trying to re-frame your first quote, really doesn't wash, I'm afraid.
You mention signals in the first quote which I took to mean that you might be aware of the difference between dynamic & periodic test signals & the fact that certain signals might reveal performance issues in a device that could be invisible otherwise, hence my post along these lines but I see that it's a waste of time.

 

[Phelonious Ponk]

No need - example already given THD 0.0015% & sound was "hard & bright". Modifications removed this issue.

So you discovered the source of the hard and bright and managed to fix it. What was the problem?

And just so I have a frame of reference, do you consider the Benchmark DAC "hard and bright?"

Tim

 

[Groucho]

I've already given you a real-world example which you reject.

Your logic really is flawed - you will accept that a battery of tests are needed to characterise an amplifier's transfer function but reduce this to only two for a DAC's transfer function. Can you please explain your logic?

I didn't say that every DAC could be perfectly described with two measurements (I'm taking it as a given that SNR and FR are very good* too), but that by studying its design one could maybe get a feel of how much trust to place in those measurements. If the internal workings of the DAC are understandable, almost mathematical, perhaps, and its load is easy and predictable, then maybe one can determine that there is no reason to believe that it cannot be adequately described by those tests. The measurements are merely confirming the theory and supporting the implicit assumption that the manufacturer has followed the data sheet.

Amplifiers are another kettle of fish. The speaker load will be unique for each type of speaker, parameters will change as temperature varies. A test that shows perfect stability into a resistor or a dummy load may not reflect reality. I accept this, and would seek reassurance that the manufacturer knows what they are talking about. I have read a lot about Quad, and I tend to trust their judgement on these matters, for example.

 

[Groucho]

Your second quoted post trying to re-frame your first quote, really doesn't wash, I'm afraid.
You mention signals in the first quote which I took to mean that you might be aware of the difference between dynamic & periodic test signals & the fact that certain signals might reveal performance issues in a device that could be invisible otherwise, hence my post along these lines but I see that it's a waste of time.

Sorry. I just mean different circumstances. Indeed, a marginal op amp (for whatever reason) may show its instability due to a change in signal. If the test was carried out at -6dB, it doesn't prove that a poorly implemented output buffer won't oscillate at -3dB. If the test used steady test tones, it doesn't prove that the output buffer won't oscillate on transients. Can I be any more open about that?

 

[jkeny]

Yes. If I am convinced by the science that says no one can prevent themselves from being influenced by extraneous factors in sighted tests, then I 'know' that I cannot audition DACs meaningfully without a blind test. This just isn't going to happen for the reasons discussed. At the same time, if I understand something about the design of various DACs, then I can make an assessment of the meaningfulness of measurements of those DACs. For example, while I may doubt that THD and IMD measurements are a complete description of an amplifier's performance, maybe they are for DACs. If all DACs boast very, very good performance figures in real world tests, then maybe I can eliminate them from my enquiries - they're all going to sound exactly the same.

The clever part of a DAC is in the silicon chip inside the tiny plastic $10 package at the heart of the 100kg machined Tungsten enclosure. If I can reasonably establish that the enclosure, power supply etc. are ridiculous affectations then I can settle on a much cheaper DAC and spend the money on better speakers.

In summary: there is no way that I am ever going to be able to audition any equipment blind, nor would I want to. So I am looking for any short cuts that can cut out areas of wasted effort/money/time. Studying measurements and design is one way to do that.

I didn't say that every DAC could be perfectly described with two measurements (I'm taking it as a given that SNR and FR are very good* too), but that by studying its design one could maybe get a feel of how much trust to place in those measurements. If the internal workings of the DAC are understandable, almost mathematical, perhaps, and its load is easy and predictable, then maybe one can determine that there is no reason to believe that it cannot be adequately described by those tests. The measurements are merely confirming the theory and supporting the implicit assumption that the manufacturer has followed the data sheet.

Amplifiers are another kettle of fish. The speaker load will be unique for each type of speaker, parameters will change as temperature varies. A test that shows perfect stability into a resistor or a dummy load may not reflect reality. I accept this, and would seek reassurance that the manufacturer knows what they are talking about. I have read a lot about Quad, and I tend to trust their judgement on these matters, for example.

Again, your second quoted post trying to reframe your first quote, doesn't wash, I'm afraid. That's exactly what you said.
If you constantly try to change what you said theres no point in discussion with you!
You are basing your evaluation of DAC's performance on these measurements as you admitted & came to the conclusion that they can be eliminated for consideration. Don't try to now use the term "perfectly described" as a get out!

 

[Groucho]

Again, your second quoted post trying to reframe your first quote, doesn't wash, I'm afraid. That's exactly what you said.
If you constantly try to change what you said theres no point in discussion with you!

Those two quotes are perfectly compatible with each other! There are no certainties, hence my use of the words "if", "may" and "maybe". Even your highlighting failed to miss one "may" and a "maybe".

 

[jkeny]

So you discovered the source of the hard and bright and managed to fix it. What was the problem?

And just so I have a frame of reference, do you consider the Benchmark DAC "hard and bright?"

Tim

Tim, it's already posted a long while back - it's John Westlake's experience, not mine. Some people called him "some dude", perhaps they should look up the guys work for a frame of reference.

Haven't heard the Benchmark so can't comment!

 

[jkeny]

Those two quotes are perfectly compatible with each other! There are no certainties, hence my use of the words "if", "may" and "maybe". Even your highlighting failed to miss one "may" and a "maybe".

Yes, the get-out, politicians words, I did not miss.
But you revealed your hand by then going on to argue for the case that DACs really don't matter as they sound "pretty much" the same.
I refuse to debate with politicians - they tend not to be interested in the truth!

 

[Groucho]

Yes, the get-out, politicians words, I did not miss.
But you revealed your hand by then going on to argue for the case that DACs really don't matter as they sound "pretty much" the same.
I refuse to debate with politicians - they tend not to be interested in the truth!

All I am saying is that until shown otherwise, I have no reason to believe that DACs that measure well don't sound good too. Edit: And they can be as cheap as you like as long as implemented correctly**.

I'm still waiting for the special engineering insight that says "THD and IMD don't reflect the quality of a DAC because...".

Just saying "DAC X sounds bad in a sighted test" doesn't get us very far. Presumably with your special knowledge you can say that "This certain DAC topology gives good measurements but fails on transients because of X, and it isn't necessary to listen to it to understand why. The people at Analog Devices/Cirrus/TI* etc. etc. didn't understand that this feedback arrangement would give high distortion when the signal did...".

That sort of thing.

* Just hypothetical examples. Absolutely no implication that any of those companies are not 100% excellent.
** as per the chip manufacturer's data sheet.

 

[Groucho]

a 500W power supply with 1F of capacitance is not going to make any difference at all to the sound, or the measurements.

Even I thought I was exaggerating a bit. But Googling brings this up pretty quickly:



(It's a DAC)

If you want high-end performance you need a high-end power supply.

Inside the NWD is a huge transformer, coupled to many thousands of microfarads of capacitance, Linear Technologies regulators and high speed, low noise diodes. We have taken every effort to make the NWD as good as it gets through careful power supply design.

It is here, in this all important power supply region, where voices and instruments take on what is known as proper “bloom” and give us the illusion of being right in the room with us. The last thing you want is an anemic sounding musical presentation – but with the right power supply, you get the fulness of the music in all its glory.

It's just such a great example of how the very existence of something makes it 'real'. Anyone without much technical background would see that and assume that some sort of engineering necessity had produced that 'solution' to what must obviously be a major problem. But such a thing can come about by other means. As a bit of an engineering type myself, I know that I could build a power supply like that by ordering about $50 of components from a catalogue. It would work fine. It would be impressive to my friends especially if I told them it was a major breakthrough and solution to a problem that plagued their own wimpy power supplies. It would almost certainly act as a self-administered placebo by making a mysterious and inscrutable $10 IC more closely resemble the look and feel of a high quality record deck or reel-to-reel tape recorder. But it would make no difference to the sound, nor the measurements.

 

[j_j]

J_J,

I have no reason to doubt of it. Are the conditions, data and results available to scrutiny of WBF members?

"Well-run DBT" is such a misused term that unless we are given good descriptive examples to debate we will go on debating futility forever.

Some are, some aren't available.

A well run test, in addition to controls, equipment verification, etc, has training, comfortable listeners, and a bunch of other things. Running a good test is not particularly easy.

I don't have any references handy.

The MPEG-2 tests were pretty good (so were the first ones, with the exception of how many people were in the listening room at once).

Things like MUSHRA I am not fond of, they attempt to lump many variables under a one-dimensional judgement. Such is not a good way to get subjects to be reliable or accurate, even individually.

 

[j_j]

I've already given you a real-world example which you reject.

And I think we'd like to see the DBT that backs up your so-called "real-world" example.

You've entered the realm of science, and now there needs to be some rigor here.

 

[j_j]

Yes, the get-out, politicians words, I did not miss.
But you revealed your hand by then going on to argue for the case that DACs really don't matter as they sound "pretty much" the same.
I refuse to debate with politicians - they tend not to be interested in the truth!

Wow, do you really thing that personal insults are going to resolve anything?

As to DAC's, I will cheerfully say that good measurements will suffice.

Now ask me how many DAC's don't do as well as expected with good measurement techniques.

Using anharmonic signals with a common difference stride can show some very amusing results, just sayin'

 

[mep]

First, I have participated in rather too many well-run DBT's.

Your second question is out of context. Use what you want when you buy something. That's not science.

As to your dishonest straw man, based on an appeal to emotion, an appeal to ignorance (disguised), appeal to facts not in evidence, and then another straw man or three, coupled with an extraction from contect and a confutation of scientific concerns with purchasing concerns (the last of which some so-called objectivists are also guilty of, but not all, and by implying universality, you widen your attack as a further emotional appeal that is both abusive and malicious), well, I think that's just a start on the fallacious nature of your insults.

DBT's are for science.

It's that simple.

When you claim an effect actually exists, you are making a testable claim, and that puts your claim in the realm of science.

If you say "I like this", that's all there is to it. You like it. Fine. Preference is inviolate.

You have completely taken everything I said out of context. My point is that I doubt many people on this forum have been involved in double-blind testing for audio which is why I asked the first question. And if you haven't been involved in audio double-blind testing, why would you constantly bring DBT up and hang it over other people's heads and ears who buy audio gear the way most of us do which is with our ears? Incessantly professing your love for something you are never really involved in is a bit silly to me. It's like having a "secret" girlfriend that you constantly tell people about and she is much prettier than everyone's real life girl friends and you take every chance to tell them that. But the reality is you really don't have a girlfriend, you just like to pretend.

To take this one step further, why except for the sake of argument are DBTs even discussed on this forum? Oh, that's right, it is for the sake of argument. It damn sure isn't because so many people on this forum are actively participating in DBTs on a routine basis and they have lots of stories to tell us about those tests. It's not like they have just read the latest copy of Double-Blind Audio that they borrowed from their "friend" with the secret girlfriend who keeps it on his nightstand next to his blow-up doll girlfriend.

If anything is a strawman argument here sir, it's the incessant talk about DBT as it relates to audio. Me asking how many people on this forum participate in DBT for audio is not a strawman argument. It's a valid question to try and put this love of DBTs in context with regards to those who profess to love them vice those who actually have any valid experience with DBTs that would pass scientific muster and the results could be published and withstand scrutiny. Based on the fact that many people who profess a deep love for DBTs of audio gear have so little real involvement with audio DBTs, they are the ones that should be accused of the dreaded strawman argument every time they bring them up and say that "we" couldn't tell "X" from "Y" in a DBT.

 

[jkeny]

And I think we'd like to see the DBT that backs up your so-called "real-world" example.

You've entered the realm of science, and now there needs to be some rigor here.

As I said it's a real-world example, not an academic exercise.
If every DAC designer had to do rigorous DBTs to evaluate every change they make progress would be kinda slow. Instead they use measurements & their ears & the marketplace evaluates the sound on a a take it or leave it basis. Does it prove anything? No. Is it the way the world works? Yes

Those seeking proof really need to learn how to live with uncertainty in their lives - this ceratinty seems to be craved by certain groups in audio forums.