Dac chips - some kind request of advice

[ginetto61]

Hi !
sorry for this question a little confused, as i am now.
Please move it to Lounge in case.
Subject are dac chips.
Let me explain the origin of my confusion.
On some audiophile discussion forums some dac chips of the past and of the present are the subject of a kind of cult.
One for instance being the PCM63. Dacs with these cult chip are very wanted.
My trivial question is ... how much the actual dac chip used can influence the overall sound out of a converter ?

Another case.

There is an audio interface very well judged by professionals, the Prism Audio Orpheus. They say that is very very good sounding in both AD and DA duties.
And also very expensive, but if the sound is good.
It uses as DAC chip (i am only interested in a dac solution) the cheap CS4398
Then looking on ebay.com i see this kit here

New CS8416 CS4398 24bit 192K Coaxial DAC Decoder Board Assembled | eBay

I wonder how much of the ... Orpheus DA sound can it provide ?
Nothing ? very little ? quite a lot ?
I can guess your answer ... buy it and try it.
Yes but is it a good way to select a kit looking at the dac chip used ?
Have you any experience of similar basic kit with a surprisingly nice sound ?
Thanks a lot for any advice.
Kindest regards, gino

 

[abraxalito]

The chip used is a very big part of the overall solution. Think of DACs (the boxes, not the chips) like the ingredients for cooking - if you don't cook with the freshest you will have to work to cover up imperfections in other ways. A decent DAC chip (i.e. a multibit one) is a pre-requisite for a satisfying sound but in itself doesn't guarantee greatness. In other words, a good chip is necessary but not sufficient to deliver a good DAC.

So you want to emulate the sound of the Orpheus or you want to go for the very best sound?

 

[ginetto61]

The chip used is a very big part of the overall solution. Think of DACs (the boxes, not the chips) like the ingredients for cooking - if you don't cook with the freshest you will have to work to cover up imperfections in other ways.
A decent DAC chip (i.e. a multibit one) is a pre-requisite for a satisfying sound but in itself doesn't guarantee greatness. In other words, a good chip is necessary but not sufficient to deliver a good DAC.

Hi and thank you sincerely for the very kind and valuable advice.
So it is important but not enough ... good !

So you want to emulate the sound of the Orpheus or you want to go for the very best sound?

Well ... given that i have never listened to the Oprheus .. let me think
I want to go for the very best sound !
Seriously ... just a musical sound would be more than enough for me
Flowing, not fatiguing, not hard or harsh ... like analog in the end.
Do you have any kind suggestion ?
I have to rely on a kit and if smd it is a big problem for me to solder. Indeed.
Both usb or spdif or optical input would be just fine.
I have a source solution already.
Thanks a lot again. Kind regards, gino

 

[abraxalito]

Take a look at the DAC on my blog (most recent post here: http://www.diyaudio.com/forums/blogs/abraxalito/1204-tda1387-octal-dac-looks-like-good-modding-base.html ).

In stock form this DAC is OK, a bit fatiguing with orchestral music. But invest an hour or so and a few $ on parts and it loses its fatiguing quality and becomes fairly satisfying. Not the last word of course but its only a $60 DAC here in China (a bit more for those who need international shipping).

Here's the thread where mods for this DAC are tried and discussed - http://www.diyaudio.com/forums/digi...-check-its-design-mod-not-play-music-not.html

 

[ginetto61]

Thanks a lot again for the very valuable advice.
Very nicely built indeed. Beautiful. Does it give some depth to the sound ?
I have noticed that the 3D soundstage effect is very challenging for most digital.
By the way ... do you have also a one/two dac chips alternative to recommend ?
something even more basic ?
Kind regards, gino

 

[abraxalito]

Yes, you put your finger on the precise thing I modded it to obtain - depth. In stock form the illusion of depth is fleeting. With the mods, it has it.

 

[Sangram]

Heard extensively the Orpheus, E-Mu 1212m, Xonar DX and a very similar little USB DAC. All with CS4398. None of them sound similar to each other.

DAC is not identifiable as being similar. The Orpheus is stellar, the cheap little USB quite bad. Not something I could live with. I own both the Xonar DX and the E-mu 1212m, and the latter sounds much nicer. Not at the level of the better AKM or Sabre chips, but still quite all right.

I don't think the DAC matters very much at at all unless the architecture optimises all aspects of performance for that particular chip. It is possible to get a very nice sound from even very basic chips (Personally speaking the Audiosector TDA1543 is a good example of a nice DAC with a very cheap chip) and it is possible to have a very good chip and still get mediocre sound. I will resist from naming examples of the latter.

 

[ginetto61]

Yes, you put your finger on the precise thing I modded it to obtain - depth. In stock form the illusion of depth is fleeting. With the mods, it has it.

Hi again and thanks a lot.
I think that digital can be divided in two ... the one that gets the soundstage depth right (ok also "tempo" is important) and the other ...
I will read your blog with the greatest interest in the weekend
Kind regards, gino

 

[ginetto61]

Heard extensively the Orpheus, E-Mu 1212m, Xonar DX and a very similar little USB DAC. All with CS4398. None of them sound similar to each other. DAC is not identifiable as being similar.
The Orpheus is stellar, the cheap little USB quite bad. Not something I could live with. I own both the Xonar DX and the E-mu 1212m, and the latter sounds much nicer.
Not at the level of the better AKM or Sabre chips, but still quite all right.
I don't think the DAC matters very much at at all unless the architecture optimises all aspects of performance for that particular chip. It is possible to get a very nice sound from even very basic chips (Personally speaking the Audiosector TDA1543 is a good example of a nice DAC with a very cheap chip) and it is possible to have a very good chip and still get mediocre sound. I will resist from naming examples of the latter.

Hi and thanks a lot for the very helpful advice.
I guess this but now i have the confirmation. The dac is just one element of the whole.
Thanks again. Kind regards, gino

 

[abraxalito]

Here's a link to a chat with a DAC designer who appears to design by listening (rather than by numbers) Bits don’t matter | audiofi.net

Key takeaway for me, paraphrased - you can get good sound out of any DAC chip provided you work at it. But sometimes you hit a roadblock you can't circumvent with the DAC chip itself.

 

[terranigma]

Here's a link to a chat with a DAC designer who appears to design by listening (rather than by numbers) Bits don’t matter | audiofi.net

This is quite confusing..he says at one passage as

The number of bits is nothing more than a marketing gimmick. No matter how many bits you have, you can’t hear beyond 16 bits. In fact, to achieve 14-bit resolution is considered very good.

How many bits you have doesn’t matter as what really matters is the effective number of bits. Unless your power supply is soooooo quiet, it’s not likely you could even achieve 16-bit resolution.

But adds:

What’s more important is the sampling frequency. I’m talking about native sampling frequency here, not the manipulated up-sampling stuff … meaning I’m talking about the way the music is converted from analogue to digital. If it’s sampled at 44.1khz (the CD standard), there is so much limitation with this format. However, if it is sampled at 88.2kHz, or 96kHz or higher, then we are talking about real music!

Music just sounds so “right” if sampled at higher sampling rates. If possible, take a 24-bit/88kHz track and down-sample to 24-bit/44.1kHz and compare it with say, 16-bit/88kHz, tell me which you prefer.

As far as I know that we can't do beyond of 48Khz with 16bits..so what is the point with mentioning about 88Khz or 96Khz whilst noticing importance of effective bits without up-sampling?

 

[Sangram]

Sampling rate and bit depth are totally different and no, this

As far as I know that we can't do beyond of 48Khz with 16bits

is incorrect.

You would be correct if you look at most popular formats and the ones in use, so there are no 16/88 music floating around.

But it is and has always been possible to use higher samplerate with lower bit depth.

Look at this chip, for eg: http://pdf1.alldatasheet.com/datasheet-pdf/view/130344/SONY/CX20152.html

 

[abraxalito]

This is quite confusing..

Yes, I don't know what he's basing his claim of 'we can't hear more than 16bits' on. I think he might be implying that for a given audio system the background noise level is hardly going to be lower than 20dBA and 16bits offers 93dB above that, so 113dB peaks. I'm sure some people's systems can go louder than that though (probably not mine!). I think assuming the baseline is the same as the background noise level also needs some justification, so it doesn't look at all a well supported claim.

As far as I know that we can't do beyond of 48Khz with 16bits..so what is the point with mentioning about 88Khz or 96Khz whilst noticing importance of effective bits without up-sampling?

As regards recording at high sample rates its a matter of whether we can hear the phase distortion inevitably introduced by the 20kHz brick wall AAF when running at 44.1kHz. If that phase deviation is audible, then the solution is to go up to 88.2kHz.

 

[Julf]

Yes, I don't know what he's basing his claim of 'we can't hear more than 16bits' on.

Have you come across any commercial recordings with a SNR over 90 dB?

As regards recording at high sample rates its a matter of whether we can hear the phase distortion inevitably introduced by the 20kHz brick wall AAF when running at 44.1kHz. If that phase deviation is audible, then the solution is to go up to 88.2kHz.

Even 48 kHz gives a lot of space to avoid really steep brickwalls.

 

[abraxalito]

Have you come across any commercial recordings with a SNR over 90 dB?

Perhaps unknowingly. I'm not in the habit of measuring my recordings, preferring to listen to them.

 

[Julf]

Perhaps unknowingly. I'm not in the habit of measuring my recordings, preferring to listen to them.

To quote Lord Kelvin: "To measure is to know".

 

[abraxalito]

Having the title 'Lord' doesn't make a person's pronouncements infallible.

 

[Julf]

Having the title 'Lord' doesn't make a person's pronouncements infallible.

No, but it doesn't automatically imply that he is wrong either...

 

[Sangram]

Have you come across any commercial recordings with a SNR over 90 dB?

Chicken/Egg?

CD-recordings do not contain information below -96dB. Think about that for a minute and work out why this is so.

Two screenshots, one is from CD and another is from DSD file. Second has information below -110dB.

 

[Julf]

CD-recordings do not contain information below -96dB. Think about that for a minute and work out why this is so.

Even better, think about it for two minutes and realize the error in your logic.

Two screenshots, one is from CD and another is from DSD file. Second has information below -110dB.

Information - or HF noise?

What is the actual SNR of that DSD file?