Delta-sigma vs “Multibit”: what’s the big deal?

[Bluespower]

why's that?

All dacs that are not multibit with no oversampling have intersample distorsion due to the fact that when you oversample the level is getting higher and then some distortion appear. Modern music is mastered near max level so very often after oversampling level is highee than the max that leads to saturate. If you plug cd player via spdif you cannot lower the level digitally before the dac. Via usb software can lower the level before dac. With digital volume on dac you can lower the volume directly and avoid the intersample distortion even with a CD player connecter on spdif. I would set digital volume to max -3db on such dacs.

 

[andreasmaaan]

I wanted to add that with dacs that don't have digital volume you cannot avoid intersample distortion from spdif. (Except some dacs that have headroom). Only nos dac doesn't have intersample distortion. Then wich is better ? No intersample distortion or no imd distortion? On usb it's obvious the difference when you set -3db in foobar on record loudly mastered.

You’re confusing things here. IMD caused by imaging is an inherent quality of NOS.

Intersample clipping only occurs in specific ocersampling DACs that have a flawed design.

There is one way to a avoid both intersample clipping and imaging IMD, however...

 

[Bluespower]

You’re confusing things here. IMD caused by imaging is an inherent quality of NOS.

Intersample clipping only occurs in specific ocersampling DACs that have a flawed design.

There is one way to a avoid both intersample clipping and imaging IMD, however...

Dx3 pro
do you think asynchronous is better than synchronous for ess dacs ?
On the dual es9038q2m from smsl topping it's asynchronous.
pro ject seems the only one to work in synchronous. In android usually i don't lile the sound cause all is converted to 48khz. That is a bad exemple for asynchronous. Maybe there is no difference for ess. In qobuzz review they say they prefer synchronous oversampling. One thing strange is that all people said the es9018 was top notch at first and now that there are the new es9028 and es9038 they say the digital glare and harshness of es9018 is gone. Why nobody told hat before the new ess? Maybe in two years people will say the same for es9028 and es9038.

 

[Veri]

One thing strange is that all people said the es9018 was top notch at first and now that there are the new es9028 and es9038 they say the digital glare and harshness of es9018 is gone. Why nobody told hat before the new ess? Maybe in two years people will say the same for es9028 and es9038.

Flavor of the month... + this shows exactly why subjective opinions are worth.. very little.

 

[JJB70]

It is a time honoured tradition of reviewing that whenever a new product is released reviewers suddenly discover all sorts of issues with the outgoing product which previously received glowing reviews. That isn't even a hi-fi thing, it is a universal thing.

I don't go beyond RBCD standard as I think it is beyond the point of no further increases in sampling rate and bit size offering an audible difference. Given that DACs are such a mature technology and given that DACs made over 20 years ago offered transparency I am a bit of a sceptic on the idea that DACs make much difference (I fully accept that they measure differently). I can readily discern significant differences in the sonic signature of speakers and headphones, when it comes to CD players, DACs and amplifiers I think very marginal differences are blown out of proportion by reviewers and most are transparent.
For all that there is a pleasure in owning something which has been superbly engineered and manufactured to a very high standard (which excludes a lot of audiophile brands) and I would like something like an Accuphase setup. My early 90's Sony ES gear wasn't cheap (although now it would be derided as mid-fi) but it was beautifully made and has given me well over 20 years of pleasure. Even though I could have paid a lot less I do not regret the money I spent.

 

[DonH56]

A quote:
Instead DSD leads to constant high levels of noise at these frequencies. The dynamic range of DSD decreases quickly at frequencies over 20 kHz due to the use of strong noise shaping techniques which push the noise out of the audio band resulting in a rising noise floor just above 20 kHz. The dynamic range of PCM, on the other hand, is the same at all frequencies.

That means that good resolution below 20khz is articial from strong noise shaping techniques. Then if noise rise above 20khz it's like a non oversampling dac. But filters in dsd are usually way above 20khz. I guess some people hear noise shaping effect and that also this hf stuff can interact with amp and speakers and resonnate at hearable levels.

The noise-shaping corner depends upon the oversampling ratio and type of noise filtering. See e.g. https://www.audiosciencereview.com/...igma-delta-digital-audio-converters-dac.1928/ for a description with some pictures of simple noise-shaping architectures and performance. The noise rises but where it becomes significant depends upon the sampling rate, and where the filters cut in. There are several other articles in the technical section about sampling, jitter, and data converter operation that may be helpful. Note that the focus is on quantization noise, one of many noise sources, and noise decorrelation (dither) is used and in general includes in-band added noise for any (multi-bit or delta-sigma) architecture.

Both multibit and delta-sigma DACs require an anti-imaging filter at the output to suppress high-frequency noise and images. Oversampling moves the images higher in frequency where they are more easily filtered. Oversampling works for conventional multibit designs as well, but the benefits are much less, i.e. only 1/2 bit for each doubling in sampling rate, vs. ~n.5 bits for an n-order delta-sigma loop, plus another m bits for an m-bit delta sigma loop. Which reminds me that most if not all modern delta-sigma audio converters include multibit converters in the loop. There is nothing artificial about the process; it moves quantization noise out of the signal band, leaving only the signal itself remaining.

HF "stuff" happens with multibit converters too, and must be filtered. See e.g. https://www.audiosciencereview.com/...ital-audio-converters-dacs-fundamentals.1927/ If it is not filtered, in any architecture, it can lead to modulation in other components or the speakers that results in audible artifacts, true. This is IME a bigger problem with the current craze of unfiltered NOS DACs that claim to sound better. So many problems with those designs not worth trying to explain.

I have not read the article but everything I have read here about it places it in yet another marketing-type paper meant to use misapplication, misdirection, and exaggeration from a germ of truth to sway buyers to their point of view.

IME, IMO, FWIWFM, my 0.000001 cent (microcent), etc. - Don

 

[Frank Dernie]

Some people prefer good sounding low fi than bad sounding hifi

I think this is the key thing.
Some products have very high levels of spurious output and/or frequency response anomalies and reading the hifi magazines a large percentage of the subjective reviewers show a preference for them.
A lot of low-fi is reassuringly expensive too

 

[DonH56]

I wanted to add that with dacs that don't have digital volume you cannot avoid intersample distortion from spdif. (Except some dacs that have headroom). Only nos dac doesn't have intersample distortion. Then wich is better ? No intersample distortion or no imd distortion? On usb it's obvious the difference when you set -3db in foobar on record loudly mastered.

All DACs can suffer from intersample distortion and from IMD (and THD, and noise, and a myriad of other distortion sources). They all create images, too, whether NOS, delta-sigma, or whatever (there are many different types of NOS DACs, for example, but all create images at the output). The amount and type of distortion and noise depends upon the architecture and implementation. Intersample distortion is not a byproduct of S/PDIF, by the way, it can occur no matter the digital interface (S/PDIF, AES, TOSLINK, USB, HDMI, whatever). It is a function of the incoming bits to the DAC itself and the low-level design of the DAC and support circuits.

It seems like this is another of those topics where folk have already decided the "truth" irrespective of any scientific basis for those opinions.

 

[Bluespower]

I think this is the key thing.
Some products have very high levels of spurious output and/or frequency response anomalies and reading the hifi magazines a large percentage of the subjective reviewers show a preference for them.
A lot of low-fi is reassuringly expensive too

I saw some studies that low fin sound of tube where good for humor. People hearing music with rich harmonics of tubes were less subject to depression.

 

[Bluespower]

All DACs can suffer from intersample distortion and from IMD (and THD, and noise, and a myriad of other distortion sources). They all create images, too, whether NOS, delta-sigma, or whatever (there are many different types of NOS DACs, for example, but all create images at the output). The amount and type of distortion and noise depends upon the architecture and implementation. Intersample distortion is not a byproduct of S/PDIF, by the way, it can occur no matter the digital interface (S/PDIF, AES, TOSLINK, USB, HDMI, whatever). It is a function of the incoming bits to the DAC itself and the low-level design of the DAC and support circuits.

It seems like this is another of those topics where folk have already decided the "truth" irrespective of any scientific basis for those opinions.

With usb you can lower the level by softwae. With a cd player connected by spdif you cannot change the level. This is true no?

 

[DonH56]

With usb you can lower the level by softwae. With a cd player connected by spdif you cannot change the level. This is true no?

Depends upon the CD player, the software, and how the circuits and processing leading up to the actual DAC are implemented. Whether from USB or S/PDIF it is serial bit stream sent to the DAC for decoding. The bits at the DAC itself are the same if the source file is the same no matter the interface. And whether from USB, S/PDIF, or whatever you can insert processing (software or firmware) to alter the bits to control the volume. That is independent of the interface. You can alter the level digitally before you send the bits (via USB or S/PDIF or ...) or after they are received. That is an implementation choice.

 

[Jimster480]

Well for delta sigma'it's much more difficult to render micro details and high details at the same,time. The test only shows that it can render them separatly. No records have song with levels at -10db or -20db so it shows nothing. But on music you can have harmonics at -40db while the global,level is at -2db.

Its not difficult at all.
Did you not read my post 3+ pages back? The DAC is literally running at 1000x+ the sample rate.
It has far more ability to render and clean up the output vs any R2R design.

 

[Bluespower]

Its not difficult at all.
Did you not read my post 3+ pages back? The DAC is literally running at 1000x+ the sample rate.
It has far more ability to render and clean up the output vs any R2R design.

R2R dacs have far more ability to be precise as they just receive the 16bits and leave them untouched. That is real bit perfect.
This article explain why delta sigma is less than 16 bit resolution in reality.

http://www.mother-of-tone.com/conversion.htm

Some quote:

: sigma-delta DACs are coarse noise-generators and when measured the way they should be measured they never make it to 16-bit resolution, don't even think about 24 bits.

1000x frequency doesn't mean same precision as 16 bit. 16 bit means 65536 levels. So to have bit perfect same precision at 20khz you should have frequency of 20 000 * 65 536 with one bit dac. I think noise shaping and dithering does the trick to keep frequency lower but on multibit you don't need any trick to get the 16 bit precision of cd. So you re sure all bit are there untouched. Bits are directly exploited.

I found more articles on internet once ago but it seems i cannot find them anymore on google. It s very strange. It seems lots of people are lying and want us to believe that delta segma dacs are better with no drawbacks and don't give arguments others than simple test that are not fair. They don't make any test with real dynamic. Each topology has drawbacks. And the drawback of delta sigma is the precision. Maybe new chipset are better for that but no test can show it. We would like to see test that show differences of precision on complex signals between dacs. Also the output stage is important on the precision. But tests only show that you have no noise. They don't show if complex signal are simplified. At the end you find lot of people who have listened to lots of dacs and they always choose to keep a multibit dac. That means something is unpleasant on delta sigma and that tests don't show it. To my mind and to my hear that is due to lack of precision.
So please stop giving strange argument like 1000*frequency means same precision as multibit. It's too simple and don't proove anything.

@andreasmaaan I think the link can interest you.

 

[Jimster480]

R2R dacs have far more ability to be precise as they just receive the 16bits and leave them untouched. That is real bit perfect.
This article explain why delta sigma is less than 16 bit resolution in reality.

http://www.mother-of-tone.com/conversion.htm

Some quote:

: sigma-delta DACs are coarse noise-generators and when measured the way they should be measured they never make it to 16-bit resolution, don't even think about 24 bits.

1000x frequency doesn't mean same precision as 16 bit. 16 bit means 65536 levels. So to have bit perfect same precision at 20khz you should have frequency of 20 000 * 65 536 with one bit dac. I think noise shaping and dithering does the trick to keep frequency lower but on multibit you don't need any trick to get the 16 bit precision of cd. So you re sure all bit are there untouched. Bits are directly exploited.

I found more articles on internet once ago but it seems i cannot find them anymore on google. It s very strange. It seems lots of people are lying and want us to believe that delta segma dacs are better with no drawbacks and don't give arguments others than simple test that are not fair. They don't make any test with real dynamic. Each topology has drawbacks. And the drawback of delta sigma is the precision. Maybe new chipset are better for that but no test can show it. We would like to see test that show differences of precision on complex signals between dacs. Also the output stage is important on the precision. But tests only show that you have no noise. They don't show if complex signal are simplified. At the end you find lot of people who have listened to lots of dacs and they always choose to keep a multibit dac. That means something is unpleasant on delta sigma and that tests don't show it. To my mind and to my hear that is due to lack of precision.
So please stop giving strange argument like 1000*frequency means same precision as multibit. It's too simple and don't proove anything.

@andreasmaaan I think the link can interest you.

You simply don't understand how signal processing works.

There is no further reason to argue with you; as you seem to be incapable of understanding the basics of signal processing or really processing at all.
As has been shown time and time again, r2r dac's are simply inferior.
It was the original way to convert audio, but it has been surpassed by better technologies.

Its like saying that a steam engine is better than a gas powered one just because steam engines were first...

 

[Bluespower]

You simply don't understand how signal processing works.

There is no further reason to argue with you; as you seem to be incapable of understanding the basics of signal processing or really processing at all.
As has been shown time and time again, r2r dac's are simply inferior.
It was the original way to convert audio, but it has been surpassed by better technologies.

Its like saying that a steam engine is better than a gas powered one just because steam engines were first...

Ok now i understand how you're not objective. It's a shame.

 

[Jimster480]

Ok now i understand how you're not objective. It's a shame.

No, I am very objective. This is why I am on an objective forum.
You are just spewing nonsense for pages upon pages already, trying to prove something that has already been proven to be false.

You are deeply lost in the audiophile marketing schemes, therefore you cannot see the light.
The power of DS has already been explained to you. There is a reason that every major recording studio on the planet is using DS based ADC's.

 

[Bluespower]

No, I am very objective. This is why I am on an objective forum.
You are just spewing nonsense for pages upon pages already, trying to prove something that has already been proven to be false.

You are deeply lost in the audiophile marketing schemes, therefore you cannot see the light.
The power of DS has already been explained to you. There is a reason that every major recording studio on the planet is using DS based ADC's.

The reason for ds dacs is money. You didn't proove anything. Anyhow i respect your choice of be sure of something without any proof. Human hear is very complex and you cannot claim that simple tests with simple tones proove anything. I think it's marketing to make test to have good specs and to not makentests that show the drawbacks.

 

[Jimster480]

The reason for ds dacs is money. You didn't proove anything. Anyhow i respect your choice of be sure of something without any proof. Human hear is very complex and you cannot claim that simple tests with simple tones proove anything. I think it's marketing to make test to have good specs and to not makentests that show the drawbacks.

There is no hope for you. As I said before I am done.

This discussion is totally off-topic at this point in time and it is also going nowhere.

Go sign up and post on Head-Fi or SBAF, those forums are filled with like-minded members who will gladly accept the nonsense you spew.

 

[amirm]

R2R dacs have far more ability to be precise as they just receive the 16bits and leave them untouched.

Those bits are still digital. In a paper design, you can just imagine them getting converted to analog by magic. In real world, you would need an array of resistors (hence the name R2R) that are anything but perfect. You would need to dial out errors between them and deal with such things as variable impedance on follow on circuits. You also have to deal with glitches when all of the bits toggle over (e.g. going form 99999 to 100000). There is a reason the industry moved on from R2R to other topologies that let us get around these serious issues.

There is no more difficult topic to understand for lay audiophiles than how DAC works. Manufacturers take advantage of this situation and give you wishful thinking of some idealized architecture when it comes to R2R. Measurements lift all of that and show the DACs for what they are: inaccurate to digital samples they are told to reproduce.

To add insult to injury, R2R DACs are also more expensive to build and hence, buy.

In this forum we are driven by how audio really works as opposed to advertising and what is written by lay bloggers. Please don't keep giving us their arguments. We know them and know then to be wrong. We say that not only with real knowledge of how they work but with measurement data. Please step back and learn from all of this.

 

[Bluespower]

Well you know everyrhong

Those bits are still digital. In a paper design, you can just imagine them getting converted to analog by magic. In real world, you would need an array of resistors (hence the name R2R) that are anything but perfect. You would need to dial out errors between them and deal with such things as variable impedance on follow on circuits. You also have to deal with glitches when all of the bits toggle over (e.g. going form 99999 to 100000). There is a reason the industry moved on from R2R to other topologies that let us get around these serious issues.

There is no more difficult topic to understand for lay audiophiles than how DAC works. Manufacturers take advantage of this situation and give you wishful thinking of some idealized architecture when it comes to R2R. Measurements lift all of that and show the DACs for what they are: inaccurate to digital samples they are told to reproduce.

To add insult to injury, R2R DACs are also more expensive to build and hence, buy.

In this forum we are driven by how audio really works as opposed to advertising and what is written by lay bloggers. Please don't keep giving us their arguments. We know them and know then to be wrong. We say that not only with real knowledge of how they work but with measurement data. Please step back and learn from all of this.

Ok thank you for independance then.
My only goal here is to think on tests that are more close to realm of the music. Because i ve doubt with ds dac for that.
You have your multitone test could you make the same test but with differencies in levels (up to at least -40db) ?
Are you not willing of improving your tests?
Are you sure they show every aspect we can hear from the dac?
I know this is classic tests that has always been. But there are always been debate about them also.
I think we should consider that lot of people prefer r2r dac and try to understand why. If you don't care it means you maybe will give bad review of dac that people could enjoy better. That is a responsability. That s why i think we should try to think on new tests and see if something is better in multi bit dacs.
I don't think they like it because of marketing. When you love music if something is bad to hear you don't keep it usually.