DAC difference
Nov 26, 2021 at 1:13 PM Post #391 of 577

gregorio

Headphoneus Supremus
Joined
Feb 14, 2008
Posts
4,231
Likes
2,355
I can perfectly understand how "in theory" all DACs should sound the same, but in practice - they do not.
In theory all DACs should sound the same and in practice they do too. With a handful of pathalogical exceptions already mentioned. Of course, we don't just have to theorize about the output of a DAC, we can actually measure the output in practice!
Do DACs need some circuitry to generate the voltage that approximates the sound?
Yes, indirectly. In fact that's the whole point of a DAC. I say "indirectly" because the output of a DAC is an analogue (voltage) signal and that signal is analogous to the sound pressure waves.
This circuitry does have a different noise level that is easy to perceive (in many cases) and then the speed of the response of voltage (and current modulations).
It does indeed have a noise level. The total noise level of even cheap mediocre DACs is below -100dB and below -110dB is not uncommon. So, if you're listening at peak levels of 100dBSPL (which I would NOT recommend!), the DAC noise would be at 0dBSPL, you think "that is easy to perceive"? But there's more, in all likelihood the recording you're listening to has a noise floor of about -60dB or higher. Do you think it's "easy to perceive" noise that's 100 times below the level of the noise on the recording? It's not that difficult to test in practice.
The case in points are USB DACs (definitely the amps are there integrated, but formally they should perfectly belong to this discussion to satisfy even the strictest moderation :))
If the discussion is about DACs, how does discussing a totally different piece of equipment "perfectly belong to this discussion"?
Here is a nice reference that I hope that you will love and appreciate as scientific objectivists: https://andyaudiovault.com/100-dongles-battle-royale/
Why?
Then what I can add is that my personal experience with USB DACs is quite similar - they can be very significantly different to a point that you can feel you are listening to a different IEM (please do not start about "faulty designs).
Personal anecdotal experience is obviously not acceptable evidence here, especially when it contradicts the actual facts. The differences between IEMs are orders of magnitude larger.
My limited take is that the speed of the response to reproduce fast transients require good circuitry, as one of the bottlenecks, especially with such small footprints and limited power, and I would be curious to hear different opinions.
Speed/time is definitely not an issue. The fastest transients have a rise time of around 40-50 micro secs, 16/44.1 has a resolution of about 55 pico secs, so just under a million times "faster" than necessary. The bottleneck in this regard most definitely isn't a DAC, it's the speakers/HP drivers that have to deal the laws of physical motion, inertia for example.
More like an amusement park not a SS
Indeed and that's because some people come here, spout utter nonsense that's completely contrary to the simple scientific facts taught in middle school and then use ad hominem attacks when they're called out on it. The solution is simple: Don't try to turn SS into an amusement park!

G
 
Last edited:
Nov 26, 2021 at 1:52 PM Post #392 of 577

PhonoPhi

Headphoneus Supremus
Joined
Dec 1, 2018
Posts
1,603
Likes
3,580
Location
Ontario, Canada
The specs I’m talking about are the ability to play 16/44.1 PCM without degradation of the signal.

The next question is What kind of degradation, and how much?

PCM 16/44.1 is perfectly capable of reproducing the sound of a violin for human ears.
A pragmatic question would be - what specs then people should look at in order to distinguish what "degrades signal" and what not. Ideally, at least for a plain pragmatic approach.

Then, I accept it as a fact that DAC/amp sound is "cooked" by different manufacturers.
Trying to preempt some of the discussions "why" - perhsps for a similar reason that "connoisseurs" prefer scotch to vodka (I cannot understand it at all, but I accept and respect it), and then the central content is still the alcohol not different smokes (we have not heard about dealcoholized scotch, right?)
 
Nov 26, 2021 at 2:01 PM Post #393 of 577

PhonoPhi

Headphoneus Supremus
Joined
Dec 1, 2018
Posts
1,603
Likes
3,580
Location
Ontario, Canada
In theory all DACs should sound the same and in practice they do too. With a handful of pathalogical exceptions already mentioned. Of course, we don't just have to theorize about the output of a DAC, we can actually measure the output in practice!

Yes, indirectly. In fact that's the whole point of a DAC. I say "indirectly" because the output of a DAC is an analogue (voltage) signal and that signal is analogous to the sound pressure waves.

It does indeed have a noise level. The total noise level of even cheap mediocre DACs is below -100dB and below -110dB is not uncommon. So, if you're listening at peak levels of 100dBSPL (which I would NOT recommend!), the DAC noise would be at 0dBSPL, you think "that is easy to perceive"? But there's more, in all likelihood the recording you're listening to has a noise floor of about -60dB or higher. Do you think it's "easy to perceive" noise that's 100 times below the level of the noise on the recording? It's not that difficult to test in practice.

If the discussion is about DACs, how does discussing a totally different piece of equipment "perfectly belong to this discussion"?

Why?

Personal anecdotal experience is obviously not acceptable evidence here, especially when it contradicts the actual facts. The differences between IEMs are orders of magnitude larger.

Speed/time is definitely not an issue. The fastest transients have a rise time of around 40-50 micro secs, 16/44.1 has a resolution of about 55 pico secs, so just under a million times "faster" than necessary. The bottleneck in this regard most definitely isn't a DAC, it's the speakers/HP drivers that have to deal the laws of physical motion, inertia for example.

Indeed and that's because some people come here, spout utter nonsense that's completely contrary to the simple scientific facts taught in middle school and then use ad hominem attacks when they're called out on it. The solution is simple: Don't try to turn SS into an amusement park!

G
Great answers!

What about the background noise as an important parameter of DACs?
That leads us to a point that the circuitry that generates voltage is not that ideal.

Can you guide me to the numbers of 55 ps, please.
~10 megabit per min is under a megabit per second of data - what do I miss. Then peak to peak it would still be those 40 kHz (?)
 
Nov 26, 2021 at 2:01 PM Post #394 of 577

bigshot

Headphoneus Supremus
Joined
Nov 16, 2004
Posts
23,415
Likes
4,936
Location
A Secret Lab
A pragmatic question would be - what specs then people should look at in order to distinguish what "degrades signal" and what not. Ideally, at least for a plain pragmatic approach.

Fine. I would say that any audible difference would be an unacceptable degradation. PCM 16/44.1 is audibly transparent. Its fidelity slightly exceeds the ability of the ear to hear. So any degradation below the specs for 16/44.1 would be an unacceptable degradation. The parameters would be noise, distortion, response and timing.

Then, I accept it as a fact that DAC/amp sound is "cooked" by different manufacturers.

What makes you think that? Are you just assuming that, or do you have evidence to back it up? Every measurement I've ever seen of DACs indicates that they exceed the fidelity of PCM 16/44.1. Generally DACs have a noise floor below 110dB, distortion levels far below the threshold of audibility, a frequency response that is perfectly flat and perfectly accurate timing. If a DAC can't even meet the standards for PCM 16/44.1 how can it claim to handle high data rate audio like 24/96 too?

Amps can have varying responses if you pair them with a set of headphones with improper impedance. But we aren't counting user error as a fault of the amp, are we?

For the footnote brigade, NOS DACs have reduced high frequency content. But NOS DACs are obsolete technology. We shouldn't be counting them either.
 
Last edited:
Nov 26, 2021 at 2:11 PM Post #395 of 577

PhonoPhi

Headphoneus Supremus
Joined
Dec 1, 2018
Posts
1,603
Likes
3,580
Location
Ontario, Canada
Fine. I would say that any audible difference would be an unacceptable degradation. PCM 16/44.1 is audibly transparent. Its fidelity slightly exceeds the ability of the ear to hear. So any degradation below the specs for 16/44.1 would be an unacceptable degradation.



What makes you think that? Are you just assuming that, or do you have evidence to back it up? Every measurement I've ever seen of DACs indicates that they exceed the fidelity of PCM 16/44.1. Generally DACs have a noise floor below 110dB, distortion levels far below the threshold of audibility, accurate timing and a frequency response that is perfectly flat.
The digital data may satisfy "transparency" conditions, but what about parameters of analogue signal generation?
Do we have a decisive set of parameters?

I assume "cooking" as a "working hypothesis" that in order to be tested needs a set of well-defined parameters on transparency, which I have not seen yet and would be happy to learn and apply.

A related question - different filters on DAC chips, e.g. ESS - how do you view them through your prism of "transparency" - what would be transparent, what not?
 
Nov 26, 2021 at 2:24 PM Post #396 of 577

bigshot

Headphoneus Supremus
Joined
Nov 16, 2004
Posts
23,415
Likes
4,936
Location
A Secret Lab
The digital data may satisfy "transparency" conditions, but what about parameters of analogue signal generation?

A DAC converts digital signal to analog signal. The specs measure how accurately it does that. The measurements on DACs *ARE* measurements of the analog signal produced by the DAC.

I assume "cooking" as a "working hypothesis" that in order to be tested needs a set of well-defined parameters on transparency, which I have not seen yet and would be happy to learn and apply.

Repeating: The parameters would be noise, distortion, response and timing. PCM 16/44.1 is audibly transparent on all of those parameters. Find a DAC that can't reproduce PCM 16/44.1 accurately, and you have found a DAC that is either "cooked" or defective. Since 1985, I have never seen a DAC that couldn't do that.

A related question - different filters on DAC chips, e.g. ESS - how do you view them through your prism of "transparency" - what would be transparent, what not?

If you are filtering the digital signal in a way that alters the way it sounds, you are doing it wrong. Yes, you can screw up sound by misconfiguring your equipment, but that is user error.

DACs are designed to meet the spec of digital audio. They are designed to be transparent. I would like to find a current DAC that doesn't produce digital audio to spec. I have never run across one myself. I hear a lot of people coming up with "working theories" regarding DAC coloration without a shred of evidence to back their hypothesis up, but no concrete examples to prove it. I can't prove a negative, so the ball's in your court. Prove it.
 
Last edited:
Nov 26, 2021 at 2:46 PM Post #397 of 577

PhonoPhi

Headphoneus Supremus
Joined
Dec 1, 2018
Posts
1,603
Likes
3,580
Location
Ontario, Canada
A DAC converts digital signal to analog signal. The specs measure how accurately it does that. The measurements on DACs *ARE* measurements of the analog signal produced by the DAC.



Repeating: The parameters would be noise, distortion, response and timing. PCM 16/44.1 is audibly transparent on all of those parameters. Find a DAC that can't reproduce PCM 16/44.1 accurately, and you have found a DAC that is either "cooked" or defective. Since 1985, I have never seen a DAC that couldn't do that.



If you are filtering the digital signal in a way that alters the way it sounds, you are doing it wrong. Yes, you can screw up sound by misconfiguring your equipment, but that is user error.

DACs are designed to meet the spec of digital audio. They are designed to be transparent. I would like to find a current DAC that doesn't produce digital audio to spec. I have never run across one myself. I hear a lot of people coming up with "working theories" regarding DAC coloration without a shred of evidence to back their hypothesis up, but no concrete examples to prove it. I can't prove a negative, so the ball's in your court. Prove it.
I did ask you about specific numbers that can enable one to decisively judge "transparency".

Now, before starting to talk "balls", again a specific question to you - for the ESS chip, let's talk ES9038, in particular, for the seven preset filters - which one would be transparent and which would not - how would you judge and teach others to judge?
 
Last edited:
Nov 26, 2021 at 3:14 PM Post #398 of 577

bigshot

Headphoneus Supremus
Joined
Nov 16, 2004
Posts
23,415
Likes
4,936
Location
A Secret Lab
I did ask you about specific numbers that can enable one to decisively judge "transparency".

Oh, I'm sorry. I figured you would know the specs for standard PCM audio. It's 20Hz to 20kHz balanced throughout the range with a noise floor of -94dB and vanishingly small distortion and timing error. These specs actually exceed the ability of human ears to hear. Even a bit worse than these specs still is audibly transparent. But PCM 16/44.1 is a convenient format to use as a standard, because it is the standard.

I'm not familiar with that particular chip. I would imagine that the manual has information about which filter is to be used in a specific situation. I would imagine that more than one of them are audibly transparent, so it shouldn't be hard to find one that works. Music is pretty forgiving compared to things like white noise. But who listens to white noise on their home stereo? And with oversampling, the filtering is pushed off into the range above what human ears can hear.

These sorts of high end features are more for theoretical improvements rather than audible ones. You really don't need to adjust the filtering. A cheap Walmart DVD player with no user configurable filtering can produce audibly transparent sound. A laundry list of filters is just an invitation for people to set it wrong and get inferior results. Audiophiles love to worry about sound they can't hear though. But we have defined our goal as audibly perfect, not theoretically perfect, so that doesn't apply.
 
Last edited:
Nov 26, 2021 at 3:26 PM Post #399 of 577

PhonoPhi

Headphoneus Supremus
Joined
Dec 1, 2018
Posts
1,603
Likes
3,580
Location
Ontario, Canada
Oh, I'm sorry. I figured you would know the specs for digital audio. It's 20Hz to 20kHz balanced throughout the range with a noise floor of -94dB and vanishingly small distortion and timing error. These specs actually exceed the ability of human ears to hear. Even a bit below these specs still is audibly transparent. But PCM 16/44.1 is a convenient format to use as a standard, because it is the standard.

I'm not familiar with that particular chip. I would imagine that the manual has information about which filter is to be used in a specific situation. I would imagine that more than one of them are audibly transparent, so it shouldn't be hard to find one that works. Music is pretty forgiving compared to things like white noise. But who listens to white noise on their home stereo? And with oversampling, the filtering is pushed off into the range above what human ears can hear.

These sorts of high end features are more for theoretical improvements rather than audible ones. You really don't need to adjust the filtering. A cheap Walmart DVD player can produce audibly transparent sound. Audiophiles love to worry about sound they can't hear though. But we have defined our goal as audibly perfect, not theoretically perfect.
OK. Got it about the range.
How do you define the noise floor, dBs are relative? What about the specs of the absolute noise (microvolts) satisfying "transparency".

DAC chip filters produce audible changes, if not right away - then through the amplification chain. Then you can see where we are getting: if the DAC manufacturers themselves give different embedded options for sound rendering - where would it leave "transparency" and "non-coloured" sound?
 
Nov 26, 2021 at 3:32 PM Post #400 of 577

gregorio

Headphoneus Supremus
Joined
Feb 14, 2008
Posts
4,231
Likes
2,355
A more specific question - do you have "specs" to record a violin and to reproduce its sound?
Sure. At a moderately close listening distance a violin has a dynamic range of about 80dB and a freq response up to a max of around 18kHz. So as bigshot stated, 16/44.1 is more than enough.
What we are getting at is the DA converter right after the DAC.
What DA converter right after the DA Converter?
Each one has a different methodology. I have already shown this. An extreme of course is the tube amp after the DAC. So those are sold as tube DACs.
Yes but of course that's just marketing, there's no such thing a a tube DAC. It is as you say, actually a tube amp and we've already discussed at some length in this thread tube amps and other very rare pathological exceptions with both amps and DACs, that do not necessarily sound the same.
But the only way some members are going to agree here is if they start admitting there is a sound difference in amplifiers?
Why would we here in SS admit something which is demonstrably false? The only way we're going to agree here is if those members who don't know or can't understand the science learn some and/or do some controlled testing themselves!
If all DACs were exactly the same and all amplifiers after them (not the actual amp that powers headphones/speakers)....the line out amp, then of course all DACs would sound the same.
They obviously don't need to be exactly the same, their differences just have to be below the level of audibility. How is that so difficult to understand?
[1] What the truth here is there are all styles of methods to alter the sound. [2] Some say different chips have a personality, but I’m not even going there. [3] All I’m saying is different line out sections give a slightly different tone to the signal.
1. True.
2. Wise move!
3. Yes they do but it's not audible.
[1] There are many ways that an even, correct and complete tone can be achieved. [2] It’s this tone that people are after. [3] Some more to the warm side, some cooler, and some very much dead on.
1. Yes, there are various different DAC methodologies that deliver a very accurate tone/freq response.
2. Good.
3. That's a contradiction! Are they after "an even, correct and complete tone" or are they after a warmer or cooler tone? If they're after a warmer or cooler tone then obviously they are not after an even, correct and complete tone. Pretty much all DACs on the market have a freq/tone response within a fraction of a dB throughout the entire audible frequency range. Except for a few rare exceptions, all DACs provide "an even, correct and complete tone", not warmer or cooler tones!
[1] The warmer Line Out sections provide a more laid back rendition of the signal. [2] The cooler ones have a push in treble or upper midrange which can expose frequencies and make the whole response seem more detailed. [3] This is personality and why people buy different DACs.
1. What warmer line out sections? And what fault do they have which means their "rendition of the signal" is "laid back", rather than accurate?
2. Which DACs have a push in the treble or upper midrange? The vast majority are ruler flat within a small fraction of a dB! Please show us reliable evidence of DACs that have a push in the treble or upper mids.
3. So that's the definition of "personality" is it, a push in the treble or upper mid-range that doesn't exist?
Just because you’ve never heard such renditions of playback doesn’t mean that it doesn’t exist.
Of course it doesn't, that's why we have science! It's not about what I have or haven't heard, it's about the objective scientific facts. You do realise this this the SS forum? So equally, just because you think you have heard "such renditions" doesn't mean that they do exist.
On the contrary it’s what the whole industry is built on.
Exactly, it's built on marketing; suggestion, shills and those whose perception has been affected by these and other biases.
Just because one DAC is different from another is true response.
Again, no one is arguing that DACs aren't different, just that the differences between almost all of them are inaudible.
[1] It’s ridiculous to state that it’s all marketing because it’s not. [2] Remember too there are over-sampling DACs and non-oversampling DACs. These have a different sound.
1. What is it then that's causing people to hear a push in the treble or upper mids that doesn't exist?
2. Of course I remember, I wrote about NOS DACs in this very thread just a few pages ago. Yes, they can sound different and are one of the rare exceptions I mentioned. Bringing up a very rare and deliberately faulty design doesn't help your case any!
[1] There are also all kinds of after effects that affect the sound. [2] DSD remastering is one, which actually may be a style of oversampling?
1. Yes there are but of course that's got nothing to do the the DAC.
2. Again, of course. The whole point of remastering is to create a new master that sounds different. Why would anyone pay a mastering engineer to create a new master that's audibly identical to an older master? But of course that's an artistic mastering process, it has nothing to do with the DAC process. And DSD is a good example of the above, a significantly different methodology with an audibly identical result.
To sit here and say the every single DAC sounds exactly the same is (Absolutely) ridiculous. What you need to say is “I have never heard a difference.” Or “I have not tried a wide range of DACs, so I’m not experienced to tell if there is actually a difference.”
Why are you trying to convince me to lie? I have heard a difference between DACs and I have heard a wide range of DACs but my experience is irrelevant! Again, this is the Sound Science forum and science is not based on individuals' listening experience, it's based on reliable evidence. Namely, objective measurements and the results of controlled tests. How is it possible not to know this?
But to assume that everyone is buying all these different DACs based on looks or expectation bias, is hilarious really!
Yes it is, but they don't seem to care about the actual facts and only want to believe the marketers, shills and other suckers. So there's not much we can do except laugh at the hilariousness of it and of course call them out if they start trying to peddle their nonsense here!

G
 
Nov 26, 2021 at 3:36 PM Post #401 of 577

bigshot

Headphoneus Supremus
Joined
Nov 16, 2004
Posts
23,415
Likes
4,936
Location
A Secret Lab
The noise floor is the distance from the peak level of the signal to the noise at the quietest level. With 16/44.1 that is a distance of about 90dB. That is a bit into overkill though.

DAC filters wouldn’t have any impact on audible sound if they were properly used. I suppose you could deliberately set them so they degraded the sound, but that isn’t the intent. I don’t see any point to that feature on a consumer playback machine.
 
Last edited:
Nov 26, 2021 at 3:50 PM Post #402 of 577

Redcarmoose

Headphoneus Supremus
Joined
Dec 11, 2008
Posts
23,174
Likes
21,375
Location
.
What DA converter right after the DA Converter?
You knew what I meant, the amplifier for line out. Tube amps for the line out totally exist, I posted one a few posts back. There are a few.


I like how you picked a choose what issues to address. You never stated that all op amps sound the same, that would have been classic.

All op amps sound the same? Do they?
 
Last edited:
Nov 26, 2021 at 3:57 PM Post #403 of 577

bigshot

Headphoneus Supremus
Joined
Nov 16, 2004
Posts
23,415
Likes
4,936
Location
A Secret Lab
Tube amps for the line out totally exist, I posted one a few posts back. There are a few.
That is a dumb idea. That isn’t the proper place to add coloration. What do you run the line out to? A solid state amp? Or double up with another tube amp? Audiophoolery never ends.
 
Nov 26, 2021 at 4:11 PM Post #404 of 577

Redcarmoose

Headphoneus Supremus
Joined
Dec 11, 2008
Posts
23,174
Likes
21,375
Location
.
Your the same guy that said tubes are inferior, then I had to show you how they are still (presently) used in all the great recording studios in the world. Then laughably you diverted the topic to, “tubes in playback”.

Please!

I went and showed you the equipment list of two TOTL studios and showed you they were using tube equipment. The tube equipment does the same thing as the tube line stage in a DAC.

It’s a particular sound that has a certain tone, no big deal? Every DAC has an noticeable difference in line out character. Some more, some less.
 
Last edited:
Nov 26, 2021 at 4:34 PM Post #405 of 577

bigshot

Headphoneus Supremus
Joined
Nov 16, 2004
Posts
23,415
Likes
4,936
Location
A Secret Lab
No, I said that some tube amps sound just as good as solid state. You are defending your position so hard you don’t listen to anything anyone else says. If you opened your ears you might learn something. I’d explain my points, but they’ve fallen on deaf ears in the past.
 
Last edited:

Users who are viewing this thread

Top