NOS DAC - Marketing BS? - Page 10

Dan, I've been thinking about any ways for people to gear the differences that we're talking about. If someone was local, I could sit them down with the DA11 or Alpha DAC and do some AB/ABX work and they'd no doubt pickup on the quality sound in general, but specificity I've been trying to think about a very pointed example, maybe using audio in the public domain or creating some that highlight the issues discussed in this thread w/ NOS dacs. The problem is of course that unless you have another reference dac to compare to, it's a moot point. I'm wondering if there might be something that could be done with the source audio before it hits the dac that would cause the inadequacies of the nos to stand out less, to trick it as it were? I'm just spit-balling here, as I obviously don't know. If that isn't possible, maybe I could find some examples that one could use on a pc or mac using publicly available dsp / filters to give them a nos sound vs os sound.

For some people lacking the ability to digest the simplest of details, you can give them explanations all day, but until an either they learn to simply trust you - or (more likely) an example hits them in the face like a 2x4 they're just going to stuff their head in the sand decrying anything they can't understand as to their "ears".

Hey.

I get it. Oversampling allows the quantization noise and harmonics to be moved higher in frequency so that it is easier to filter out, keeping it from modulating down into the audible range and causing distortion. NOS requires rolling off the highs in order to get acceptable attenuation of that noise, and even then doesn't do that good a job.

Got it?

Obvious?

Obvious? Not sure if you have a chip on your shoulder, but that wasn't addressed to you.

Pardon me.

You seemed to need a clarification.

You're winning all kinds of awards today in the realm of perception

I have read this thread with great interest, Dans points are technically correct and cant really be argued with.
What is more interesting to me is the audibilty of the high frequency energy that a NOS would introduce into the signal.
A friend of mine has a 'NOS' dac and it sounds fine to me , there may be a few reasons for this, one is my hearing cuts off at around 17khz and i may not be picking up any noise up, secondly he uses a tube amp and a Klipsch horn speaker, the tube amp has probably got output transformers that filter out some noise and/or horn simply wont produce the noise at all.

It reminds me of the arguements between Grado and Sennheiser fans, most agree that Grado produce a 'coloured' sound that isnt as accurate as some of Sennheisers higher end cans but many love that colour!

astroid - you're quiet correct, not knowing exactly which horns or tube amp it is, the odds are they're doing what those designs did/do in the classic sense; introduce noise and reproduce the 'tube' sound. That sound defiantly appeals to me at times, it would be my opinion thou to insure the speakers or the amp is what is coloring the audio.

Based on what you said here, I would say you only got a part of it, and not too well. The main problems with Nos are:

1. NOS kills the flatness response.

2. Nos introduces high amplitude image energy (above hearing), which impacts the behavior of what it drives (headphone amp, power amp or what not). You are correct to say that some of that image energy modulates back to the audio range. We call it "inter-modulation" at such energy appears across the whole audio range, from 20-20KHz.

3. Some of the high frequency image energy can be filtered, but to a far lesser degree then OS. A analog filter for NOS introduces ADDITIONAL loss of high frequencies.

4. And as rule, when filtering a NOS one impacts the phase. BTW a filter at (even the simplest and weakest 1 pole) at 20KHz impacts phase all the way down to 10KHz. Very few filters have a phase compensation.

Around 2 dozen years back, all we knew is NOS, and we HAD TO live with it.
Since then, we have OS and ALL the major DA makers are doing OS and nothing else, and that is so to overcome ALL the problems mentioned above.

I am writing here the "condensed list" of what NOS does. My previous posts were for explaining the how and why, and I tried to keep it as simple as I could for the benefit of the technical readers.

At this point I can see that Benjamin is correct in saying that most people here do not understand much of what I say.

I thought that if I just come up with a list, such as in this post, it may be considered as having as much or as little value as any other claims, including wishful thinking, commercially driven and what not. So I made sure that what I say comes with a technical backing - real solid facts. But that requires a reader that is able and or willing to comprehend.

So in this post I reduced it to 4 points, and it is backed by the previous posts (for the willing and able and interested). Does anyone need even a simpler version? Here it is:

High fidelity audio, is a term made long ago, when audio gear was real poor.
Long ago, 5KHz was tough to get, and the state of the art was with much distortions and noise, low dynamics. People wanted the gear to be able to reproduce music the way it really is - with high degree of transparancey. To do that, they wanted to cover the hearing range of humans, including good ears and young ears. Testing suggested that 20Hz to 20KHz is a good goal, and Hi-Fi was born. That was even before stereo. In an ideal world, we would want a flat response with no attenuation even at 20KHz. Hi-FI was already a compromise by itself, agreeing to a 3dB loss at 20KHz. There are some people that can can hear a little above 20KHz.

Analog technology improved first, and the 20-20KHz was no longer a future dream. But then came digital, and the 44.1KHz standard along with it. In practice it was a compromise. There were reasons for the compromise - to limit the amount of data. (You did not want a CD to play for say only 20 minutes).

So we did our best, and as long as everything was NOS (that was all we knew), we encountered the problems mentioned above. But with the invention of OS, the compromises went away.

The 20-20KHz was not an arbitrary thing made by some geeks and EE's. It was done by testing what people can hear.
Do you want a camera that does not capture red? or Blue? The idea of reproduction is to be able to capture everything. You can muck with it later, such as bass boost or what not, but the starting point should include everything. NOS does not lose highs and distorts because such is a desirable outcome. It does so because of technological limitations. If you like NOS, you are accepting gear that impacts the sound by NOT reproducing the highs, and causes other problems. If that is what you like, so be it.

I am against gear that can not reproduce what the ear can hear. Audio gear should cover whatever the ear can hear and with some safety margin. You don't have to agree.

But are you willing to say that gear is better when it removes a part of the music? Any music? Does all music sound better with less highs? Or is the argument based on "being older" so one can not hear as well as they used to?

Regards
Dan Lavry
Lavry Engineering

So in the end, does this not simply boil down to the classic holy war, "measurements vs. my golden ears" ?


this thread has convinced me to find a DA10/11 to demo I feel a purchase coming...

More like measurements vs. gold plated turds. Ears are fairly predictable, what's between them is not.

NO it does not. The measurements are there to ensure that the gear meets some of the ear capabilities.

Say we know that a good ear can hear down to 120dB dynamic range. You do not need to listen to each individual gear. You can measure that.

Say we know that we want to have .1dB response from 20-20KHz. We can measure that.

Then things get more complex. We can measure precisely all sorts of complex tones and combinations...

My audio Precision system has a better absolute pitch than any human. It yields results that are very accurate and objective.

Of course, the test gear does not have the ability to test everything, Music is "anything" within the hearing range, and one can not test everything and anything. So listening is very much in order!!!

The test gear is much better and more accurate for telling you exactly what the unit under test does not do well, for a large number of conditions. Say your test gear tells you that the noise is at -60dB. Do you need an ear to tell you this is bad news? In fact, one human may guess it is at -50dB, another will call it 70dB.

The test gear is there to tell you what it sees when you "hit" the gear with many known signals. The answers it yields are much more accurate then a human, and that can be used to great advantage. It can certainly point out many shortcommings. A good and experience designer has a large number of tests, and better understanding of music and the ear makes for better testing.

But again, the testing, as complex as it may be, does not cover everything. When I said "music is anything and everything within the range of human hearing" I mean just that. Music is not just a bunch of tones. One can learn much from testing but one can not ever learn everything.

When one is done with the measurements, and everything looks fine, one listens. And when things do not sound right, a good designer will do what they can to come up with a measurement that will shed a light on the problem. That is one way for a designer to have a better "template" (a set) of tests.

So measurements have their place AND a golden ear has it's place. It is NOT one or the other. It is BOTH! At the end of the day, it is the human hearing drives the testing, and improve the methods of testing.


Measurements yield precision numbers and plots for limited tests. Listening covers a wider range (music) but the outcome is not precise. A listener can not put numbers or be too precise regarding the cause of sonics. But measurements and hearing support each other, they both have their place and they do not go against each other. It would be good if folks would realize it.

One can not go against hard numbers when they are poor. It is much easier to do some "fast talking" about "it sounds good". It is not surprising that when gear measures poorly, the makers and sellers put down measurements and tests.

If transparancey is the goal, and testing says it sucks, the ear would agree. The converse is not always true. If the gear measures well it may not always sound well, because the testing missed something that the ear can hear.

Regards
Dan Lavry
Lavry Engineering

Unless you can find an amazing deal on a used DA10 or something along those lines, I would push for the DA11 as PiC is well worth it, and green LEDs are just nice to look at In any case try and find something mastered using Lavry AD hardware, an easy find is the last Star Wars movie soundtrack; Revenge of the Sith. Something that is truly reference quality (hopefully something you will be familiar with as well) will just knock it out of the park for you. It's not to say the Jonas Brothers won't sound better (ok.. maybe not), but I'm not so sure you want to get any closer to that kind of music. The quickest way for me to tell how well a piece of kit is in the realm of transparency is to listen to recordings of my parents, wife and kids just talking - recorded using Lavry's AD10 + his preamp, or on the Pacific Microsonics 2 - as well as recordings of my wife playing viola, and my sister playing harp / harpsichord

Now that's an awesome soundsystem

The measurements I've posted were taken by me with an Audio Precision ATS-2. Everything was plugged into the same power strip. (The power strip and electrical circuit has a bunch of computer gear on it, so this was a real-world environment.) The measurements have a little note about 60Hz because I found even with nothing attached the ATS-2 was measuring 60Hz and harmonics. The D100 did not contribute any 60Hz noise.

What was used for the A/D or mastering does not have any impact on what you should use for D/A conversion, because we always create a standardized format: e.g. PCM, DSD, MLP. There's nothing left in the resulting "file" where you would change your D/A process based on how the file was created.

Using the camera analogy, your typical consumer camera creates a JPEG image. No matter what camera you used, any image software can read and display that JPEG, and they'll all display the same thing. (RAW doesn't count, because that isn't a single standard.)