Attempting to better understanding DAC testing

[DanaGer]

I swear this is a legitimate question. I’m not attempting to start another war of us vs them but am trying to truly understand the intricacies of DAC testing. I’ve read the article Understanding Digital Audio Measurements and it makes sense testing not just one but multiple frequencies to make sure they’re linear and measuring the distortion to make sure what goes in is as close as possible as to what goes out but (there’s always a but) in the attempt to reduce distortion are some of the nuances also being filtered out? If the frequencies are being generated by a tone generator and not a musical instrument are the “distortions” associated with that instrument being somehow masked in the name of providing a better number? A 261.63 Hz (middle C) played on a piano does not sound like the same frequency played on a saxophone. It actually doesn’t sound the same played on the same instrument of a different brand. Can testing identify a wave form produced by a Stradivarius violin as opposed to say a Yamaha violin. I can hear the difference between the two (If I can’t I’m sure there are people who can) and I would think it could some how be measurable. So my question is the baby being thrown out with the bathwater? Is the attempt to produce a better measuring DAC, are manufactures sacrificing nuances which would explain why some prefer the less numerically accurate sounding DACs? There are a lot of people here who know a lot more about this stuff than I do. I would appreciate the “Jane you ignorant slut” crowd to keep it to yourself but if there is an error on my logic please educate me.​

Cheers,

 

[AudiOhm]

Sorry I could not get past...because of the text size and bold...

I swear this is a legitimate question.​

Ohms

 

[DanaGer]

By holding the" CTRL and tapping -" you can reduce the screen resolution if that would help. Us older guys don't always find the fine print legible.

 

[Trell]

By holding the" CTRL and tapping -" you can reduce the screen resolution if that would help. Us older guys don't always find the fine print legible.

Your use of [ HEADING=2 ] makes other text real small then, so why not just use normal font size and let the readers decide?

 

[voodooless]

How is better distortion going to filter stuff out?

By definition, distortion is the stuff that wasn’t originally there. So if you have less of it, the more of the original is left.

 

[GXAlan]

That is your Multitone testing.

Are Yamaha products more musical than Sony products? (Quick Measurements)

Quick and simple comparison, where I'm just looking at two measurements from the pre-outs of two classic products in pure-direct modes 1) Sony STR-ZA5000ES - Similar digital front end as the STR-Z1100ES tested here but with beefier amplifier section - 90 dB ASR DAC SINAD on APx555 2) Yamaha...

www.audiosciencereview.com

On a Yamaha AVR, the 1 kHz and multitone differ by less than 1 dB.

On a Sony ES AVR, you lose 7 dB of SINAD!

 

[staticV3]

in the attempt to reduce distortion are some of the nuances also being filtered out?

The only thing being filtered out by modern DACs is DC and content above the Nyquist frequency. Both are good things that project your gear. Neither filter out any fidelity.

If the frequencies are being generated by a tone generator and not a musical instrument are the “distortions” associated with that instrument being somehow masked in the name of providing a better number?

The DAC will play a test tone as precisely and purely as it will play any instrumental recording.
The DAC does not and cannot differentiate between the two.

A 261.63 Hz (middle C) played on a piano does not sound like the same frequency played on a saxophone. It actually doesn’t sound the same played on the same instrument of a different brand.

That's called timbre, yeah.

Can testing identify a wave form produced by a Stradivarius violin as opposed to say a Yamaha violin.

Of course.

Is the attempt to produce a better measuring DAC, are manufactures sacrificing nuances which would explain why some prefer the less numerically accurate sounding DACs?

That's simply not how things work.
Optimizing a DAC for measurements does not mean it'll turn beautifully nuanced performances into test tones out of a sine gen.
It means it'll play everything, from test tones to electronica to orchestral performances with the highest fidelity possible, as true to the recording as can be.

 

[GXAlan]

Is the attempt to produce a better measuring DAC, are manufactures sacrificing nuances which would explain why some prefer the less numerically accurate sounding DACs?

This is still a million dollar question.

It is not fair to say that they are sacrificing nuances unless you see worse multitone measurements.

One area where there can be differences in practice are
1) ground loops.

Amir tests best case. In real homes, there can be big differences. I have a Topping D50s which measures worse than my Korg DS-DAC-10R because my USB power is noisy and there is some ground loop. My Marantz PM-10 measures better than my PM-11s2 but it is more sensitive to ground loops too.

So the better measuring equipment on an APx555 might measure more poorly in a real home. This is the power of running your own test gear at home.

2) There is a fallacy that better sounding DACs are more musical. They may in fact have euphonic coloration. There is a fallacy that transparency is always best.

Imagine listening to classical music where the lowest levels are lost in noise versus a system that allows you to hear the rustling of the sheet music pages being turned.

If the noise is unobtrusive, our brain is more likely to enjoy the less transparent playback system as compared to the intermittent page turning which doesn’t reflect the experience from the best seat in the house.

What about actual music content that is that quiet? The composer/conductor would not have composed or interpreted the music to be at a level so soft that no one in the audience could hear it.

Transparency to an overly detailed recording would not be as pleasant to the ear than one that hid some of those details in the noise.

 

[staticV3]

Amir tests best case

No he does not.

 

[Chrispy]

By holding the" CTRL and tapping -" you can reduce the screen resolution if that would help. Us older guys don't always find the fine print legible.

Better if you put your posts in the regular resolution and do that what you suggest when you need it. I think you're still imagining properties of electronics due too many years of reading "audiophile" nonsense.

 

[GXAlan]

No he does not.

Would help to articulate instead of a post like that.

I have done a lot of measurements myself INCLUDING on the same exact specimen that Amir has run his tests on.

1) Amir has a more controlled measurement environment than the average home user. It is harder to get the same results at home. This is what I mean by best case scenario.

2) Showing better results at another website doesn’t offer any confidence when there is potential for lot to lot variability.

3) Amir won’t “work” to generate the best case in most cases — he will test it out of the box in most cases. One example would be the JBL4319 with one click too high in the treble (it has microclicks as opposed to being truly continuous) so the measurements show the treble being hotter than it actually measures. BUT it is reflective of out of the box control and how silly the treble control is when it goes from 0 output to 100% as opposed to something like a +/- 5 dB range. His data isn’t wrong just incomplete.

4) I will say that I get close matching results to Amir comparing the same specimen of the Panasonic UB9000 while I get vastly worse results with my Topping D50s that I bought myself.

 

[Killingbeans]

If the frequencies are being generated by a tone generator and not a musical instrument are the “distortions” associated with that instrument being somehow masked in the name of providing a better number?

No. The harmonics from a musical instrument are massively higher in amplitude than the harmonic distortion you get from even a badly designed DAC.

The music is actually doing the masking. It's much easier to hear distortion of a single tone, than it is to hear the same amount of distortion in a piece of music.

 

[DanaGer]

The DAC will play a test tone as precisely and purely as it will play any instrumental recording.
The DAC does not and cannot differentiate between the two.

My wording is off then. Not the DAC, can the test results differentiate between the two The overtones and harmonics "timbre" as opposed just the frequency produced by pure tones. Are the test tones played as a chord all the same time or are they measured as a sweep? How do harmonic overtones different from harmonic distortion. I would there would be at least some interaction between the two.

 

[voodooless]

How do harmonic overtones different from harmonic distortion.

One is already there, the other is newly generated by the device

 

[antcollinet]

I swear this is a legitimate question. I’m not attempting to start another war of us vs them but am trying to truly understand the intricacies of DAC testing. I’ve read the article Understanding Digital Audio Measurements and it makes sense testing not just one but multiple frequencies to make sure they’re linear and measuring the distortion to make sure what goes in is as close as possible as to what goes out but (there’s always a but) in the attempt to reduce distortion are some of the nuances also being filtered out? If the frequencies are being generated by a tone generator and not a musical instrument are the “distortions” associated with that instrument being somehow masked in the name of providing a better number? A 261.63 Hz (middle C) played on a piano does not sound like the same frequency played on a saxophone. It actually doesn’t sound the same played on the same instrument of a different brand. Can testing identify a wave form produced by a Stradivarius violin as opposed to say a Yamaha violin. I can hear the difference between the two (If I can’t I’m sure there are people who can) and I would think it could some how be measurable. So my question is the baby being thrown out with the bathwater? Is the attempt to produce a better measuring DAC, are manufactures sacrificing nuances which would explain why some prefer the less numerically accurate sounding DACs? There are a lot of people here who know a lot more about this stuff than I do. I would appreciate the “Jane you ignorant slut” crowd to keep it to yourself but if there is an error on my logic please educate me.​

Cheers,

I think the misunderstanding you are making is in thinking that reducing distortion harmonics from the output of a dac somehow will result in the harmonics created by the instrument being filtered out.

That is not how it works. Harmonics are not removed by a good DAC, they are added by a bad one, and the added harmonics are a form of distortion - they change (distort) the shape of the waveform.

On the other hand the instrumental harmonics in the recording are safe. Why? Because the DAC has no way of telling which frequencies in music are harmonics, and which are fundamental tones from the instrument. It is simply a wavform that it is supposed to reproduce without ADDING any harmonics or noise of its own.


Think of it this way. Lets say you have a recording of a violin playing middle C (256Hz). Lets say this violin adds 2nd 3rd and 5th harmonics (in reality it will be much more complex than this, and the harmonic frequencies will all be at different amplitudes.

So the freqencies in the recording are 256Hz, 512Hz, 768Hz and 1280Hz.

Now if you put that waveform through a perfect non distorting dac - that is exactly what you will get out. The timbre of the violin will be reproduced correctly

However, if you have a distorting dac, that adds it's own significant harmonics (say 2nd and third), then every frequency in the recording will have harmonics generated, so the dac will add 2nd and third harmonics of all the frequencies in the recording - 256 plus 2nd and third - 512 plus 2nd and third - 768+ 2nd and third 1280+2nd and third. All these additional harmonics will CHANGE the sound/timbre of the violin. Worse all these frequencies will intermodulate creating a spray of non harmonic frequencies at plus/- the difference between each pair of frequencies.

Now imagine this happening on all the frequencies in a recording. All the notes with all the harmonics from all the instruments and voices. THIS is why we want reproduction equpiment to be as clean and non distorting as possible.

 

[Killingbeans]

Are the test tones played as a chord all the same time or are they measured as a sweep?

Depends on the test.

How do harmonic overtones different from harmonic distortion.

One is intentional, the other isn't.

 

[DanaGer]

One is already there, the other is newly generated by the device

Can testing differentiate between the two? What do overtones look like on the dashboard? One is considered distortion one is considered timbre can we reduce one without affecting the other? Can overtones even be measured? Even at the DAC chip level each chip produces a different measuring output. There is processing going on both pre and post of the signal conversion and the signal is being altered by the components the system is made of. The testing is performed post processing and test tones are a known value to test against. So could a test be constructed to use tones produced by analog instruments and what would results look like?

 

[antcollinet]

One is considered distortion one is considered timbre can we reduce one without affecting the other...

Read my post. The dac doesn't reduce harmonics, a good dac just doesn't add any.

 

[DVDdoug]

why some prefer the less numerically accurate sounding DACs?

Sometimes Audiophiles are called "audiophools" because they often hear things that normal people cant hear*... and that they can't actually hear in blind listening tests, and then they use nonsense terminology like "musical sounding" or "analytical" and thousands of other vague-meaningless words instead of talking about real characteristics like noise, distortion, and frequency response.

Sadly, the audiophile community is dominated by audiophools and the term "audiophile" has developed bad connotations.

Audiophoolery might be worth reading.




* There are trained listeners and others wo can hear better than the average person, just as some musicians have perfect pitch. But it's a big red flag if they aren't using precise scientific or engineering language language.

 

[DanaGer]

Read my post. The dac doesn't reduce harmonics, a good dac just doesn't add any.

I'm not sure that's true. All DACs add some distortion just like amps and speakers do. Good measuring DACs just add less. Testing shows that DACs with the same DAC chip measure differently . If the distortion isn't coming from the chip itself (same DAC chip) and one measures better than the other then something is reducing the distortion or at least adding less of it.