• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Best DAC for I2S input

Wuzel

Member
Joined
Nov 30, 2021
Messages
35
Likes
22
Out of (real time) recording sessions u chain doesn't need a world clock...
Most DACS (all these days?) reclock the signal for the sake of jitter reduction. And most of them use for these absolut high-end femto clocks.
I think it's nearly impossible to put an external world clock with the same precision in.

And - again - why u need these?
 

mete

Member
Joined
Dec 31, 2021
Messages
11
Likes
1
I believe, far from an expert here, but the only time for certain is if you use external clocks from like a ddc. It's just that I2S is supposedly sending all that information unpacked or idk what in gods name. Certain digital inputs send info that needs to be messed with prior to playing which I dont know the hardcorest of hardcore may say degrades the sound.

I honestly dont know what to say, I think it all comes down to the main components themselves like the streamer, I am using I2S and I sure as heck didn't dish out 300 dollars for a cable. Found a specific i2s cable which has exact pin out as my streamer with an hdmi equivalent. Based on the information above it's not i2s that is actually making everything sound better it is the streamer. Dont have AES to test between the two, dont care.

I am virtually done chasing anything any more, I would suggest the same unless your pockets are deep.

OK, also as far as I understand from different places, other than providing an external clock directly, a DAC is not using the clock in the I2S for DA conversion.
 

mete

Member
Joined
Dec 31, 2021
Messages
11
Likes
1
Out of (real time) recording sessions u chain doesn't need a world clock...
Most DACS (all these days?) reclock the signal for the sake of jitter reduction. And most of them use for these absolut high-end femto clocks.
I think it's nearly impossible to put an external world clock with the same precision in.

And - again - why u need these?

I dont know if I need it or not, but trying to understand. I asked about I2S because only a few and mostly pro audio gear have external clock inputs, so I wondered if it is also used in this way.

Why is it nearly impossible to put an external clock in with the same precision ? If I am not wrong, OCXOs have better short term stability than femto clocks, and OCXO clock generators are used in different industries, I assume they do not sacrifice this benefit just because they are an external component.

EDIT: I am not sure if the strike-through statement above is correct. I would like to know if you have a supporting or not supporting knowledge about it.
 
Last edited:

MacCali

Major Contributor
Joined
Dec 21, 2020
Messages
1,121
Likes
535
OK, also as far as I understand from different places, other than providing an external clock directly, a DAC is not using the clock in the I2S for DA conversion.
No clue, maybe that's what bit perfect means, yet there's other connections which are bit perfect too. Dont understand what bit perfect is actually doing. Or maybe bit perfect is produced by the audio output source.

But as I said above I believe it is optical as the example of sending information in basically "1 packet" and then the dac gets it and has to unpack the information before turning it into analog. I2S is basically sending 3 Unpacked "packets", guessing that all digital sources are sending the same 3 things that I2S sends or something very close.

So besides jitter, supposedly when you get direct from the source without having to do extra work on the internal dac it makes sounds better, but there's not measureable difference maybe it is again a time alignment or the source itself making it better
 

mete

Member
Joined
Dec 31, 2021
Messages
11
Likes
1
No clue, maybe that's what bit perfect means, yet there's other connections which are bit perfect too. Dont understand what bit perfect is actually doing. Or maybe bit perfect is produced by the audio output source.

But as I said above I believe it is optical as the example of sending information in basically "1 packet" and then the dac gets it and has to unpack the information before turning it into analog. I2S is basically sending 3 Unpacked "packets", guessing that all digital sources are sending the same 3 things that I2S sends or something very close.

So besides jitter, supposedly when you get direct from the source without having to do extra work on the internal dac it makes sounds better, but there's not measureable difference maybe it is again a time alignment or the source itself making it better
Bit perfect just means the bits in the original audio file/stream arrives to DAC without any modification. This might not be the case if the audio stack (mixer etc.) of the OS is used in computer based playback. If the transmission (SPDIF, I2S) affects the bits, there is a big electrical problem which has to be fixed, it cant be compensated.

SPDIF basically packs 16-24 bits of audio into 32 bits packets, and also embeds the clock into this (technically it means bits are encoded with differential manchester scheme) whereas I2S transfer bits directly (without a special encoding) and carries clock separately. At the end they all carry the same information but 1) I2S is easier to decode because there is no need to recover the clock as in SPDIF (but practically this does not matter much anymore), 2) I2S can carry more data (higher kHz, higher DSD bitrate) and also can carry 32 bit samples, 3) related to (1) a bit, because of encoding, the data in SPDIF affects the transmission properties (again this probably does not matter anymore). So in terms of transmission, as long as the data can be read correctly from SPDIF (ie the jitter is not enormous), there is no difference between them when neither the embedded/recovered clock of SPDIF nor the clock in I2S used in DA conversion process which is the case as far as I understand in all modern DACs.

With I2S, I think, theoretically it should be possible to connect source to DAC and DAC chip directly without any jitter reduction process (eg DPLL in ESS), but I couldnt find much information yet about the details of this. I2S is, AFAIK, a chip to chip (same board/PCB) protocol, an external connection maybe makes it not possible to feed the data and clock to DAC chip directly.
 

AudiOhm

Senior Member
Joined
Oct 27, 2020
Messages
406
Likes
409
Location
London, Ontario, Canada
In my main system I have a DV 9500 for all my disks including SACD that is only output via the analog outs.

In my second system I have a Sony UBP X800M2, unfortunately while playing SACD's it's output is PCM.
Will there be an upgrade in sound if I use a separate DAC (ex. Gustard X16) that can utilize the IIS output of the Sony?

Ohms
 

MacCali

Major Contributor
Joined
Dec 21, 2020
Messages
1,121
Likes
535
In my main system I have a DV 9500 for all my disks including SACD that is only output via the analog outs.

In my second system I have a Sony UBP X800M2, unfortunately while playing SACD's it's output is PCM.
Will there be an upgrade in sound if I use a separate DAC (ex. Gustard X16) that can utilize the IIS output of the Sony?

Ohms
If you read above the link attached, Amir did a test to see if there’s any difference using i2s. Measurements wise the conclusion is no, it doesn’t get any worse than what the DAC already can produce.

As mentioned above, no clue if the i2s signal does anything to help your dac[any dac] not have to unpack anything and provide a cleaner sound because the dac is doing less work or modifying of the signal.

Honestly CD’s themselves are to my understanding far superior to streaming. You are getting the maximum quality of music and I doubt switching to i2s will do much if anything or placebo.

Only thing, as I do not know the quality of your internal dac or the age of your CD player, I did not look and I doubt there are measurements, you will probably get an improvement if you use the Gustard, and if synergy is there that will help as well. If you got more dacs you can try those and see what works best.

Also with your Sony you need to buy an i2s converter, from what I searched online, you use audio out to an i2s converter and honestly not sure what that may create[sound degradation]

Second, most importantly since i2s is not standardized you would need to make sure the pin layout is the same as the Gustard or it won’t work.

This is why they suggest not using i2s in the first place, cause it’s unlike most digital inputs. Again this is all posted above on this thread I believe
 
Last edited:

avoidwork

Member
Joined
Dec 10, 2020
Messages
60
Likes
37
Location
Ottawa, Canada
When I2S input is used, the internal clock circuitry of DAC is not used at all ? So I2S can/should provide a better clock than internal one ?
I2S should never be used. Like AES, it's a relic... but it's also misapplied for this use case.

Use USB if you want the highest quality signals; if the cable run length requires AES that's unfortunate.

I2S is common bus for sending data, but it's not audio centric nor logical to use. It's maker friendly with SBCs, etc.
 

Roland68

Major Contributor
Joined
Jan 31, 2020
Messages
1,352
Likes
1,121
Location
Cologne, Germany
I2S should never be used. Like AES, it's a relic... but it's also misapplied for this use case.

Use USB if you want the highest quality signals; if the cable run length requires AES that's unfortunate.

I2S is common bus for sending data, but it's not audio centric nor logical to use. It's maker friendly with SBCs, etc.
I2S wasn't specially developed by Philips for the transport of digital audio data between chips?
It is not the format that is used in almost all DACs to feed the DAC chips, or to extract the corresponding right / left data?
And which data format is obtained from USB in the DAC?

Or do you confuse it with I2C?
 
Last edited:

MacCali

Major Contributor
Joined
Dec 21, 2020
Messages
1,121
Likes
535
I2S should never be used. Like AES, it's a relic... but it's also misapplied for this use case.

Use USB if you want the highest quality signals; if the cable run length requires AES that's unfortunate.

I2S is common bus for sending data, but it's not audio centric nor logical to use. It's maker friendly with SBCs, etc.
Yes we established this a while back on this thread I believe
 

MacCali

Major Contributor
Joined
Dec 21, 2020
Messages
1,121
Likes
535
I don't think there's any evidence of that.
Well we already see for certain when something is up mixed or whatever it’s called, from cd to 24/192 or any change artifacts are introduced which degrade the audio quality. Unless artifacts are a good thing, that’s not what I got from it.

This is what I understood per Amir’s testing of high res.

No clue what is being done in the streaming process to comment. But I am sure you have heard that music on a drive vs a usb stick has differences as well. So one getting access to direct data vs over your network[NAS]. It’s what people claim, maybe placebo who knows.

Maybe some day we will find out, I stream and I do not own a CD player to even test this or a NAS, or even downloaded music to test via usb.

I don’t think there’s either evidence that proves or doesn’t.
 

BDWoody

Chief Cat Herder
Moderator
Forum Donor
Joined
Jan 9, 2019
Messages
6,948
Likes
22,625
Location
Mid-Atlantic, USA. (Maryland)
But I am sure you have heard that music on a drive vs a usb stick has differences as well. So one getting access to direct data vs over your network[NAS]. It’s what people claim, maybe placebo who knows.

I haven't...

My bet is on placebo.
 

avoidwork

Member
Joined
Dec 10, 2020
Messages
60
Likes
37
Location
Ottawa, Canada
I2S wasn't specially developed by Philips for the transport of digital audio data between chips?
It is not the format that is used in almost all DACs to feed the DAC chips, or to extract the corresponding right / left data?
And which data format is obtained from USB in the DAC?

Or do you confuse it with I2C?
You're correct afaik; I2S from the USB controller (xmos xu208/216) to DAC. The critical difference is USB is standardized pin out, etc., while I2S is not. It's not the ideal transport between devices, it's meant for internal usage (in this application).
 

avoidwork

Member
Joined
Dec 10, 2020
Messages
60
Likes
37
Location
Ottawa, Canada
So besides jitter, supposedly when you get direct from the source without having to do extra work on the internal dac it makes sounds better, but there's not measureable difference maybe it is again a time alignment or the source itself making it better

Nothing is being made better. At best you're getting a 1:1 reproduction, and at worst you're dealing with clock sync issue via jitter.

An external clock signal is added complexity and added risk, it is not beneficial unless your DAC which is receiving has a really ****** internal clock, but most have femto clocks when you get into "end game".

My smsl vmv d3 even has the connection for a 10mhz clock... but it's got 2 femto inside so zero point in using it.

This might be appealing with a topping dac, but ugh, lot of effort to get away from a crackle when playing non dsd after dsd file.
 

Roland68

Major Contributor
Joined
Jan 31, 2020
Messages
1,352
Likes
1,121
Location
Cologne, Germany
You're correct afaik; I2S from the USB controller (xmos xu208/216) to DAC. The critical difference is USB is standardized pin out, etc., while I2S is not. It's not the ideal transport between devices, it's meant for internal usage (in this application).
Yes and no.
I2S is absolutely standardized with an option for the master clock, but we're talking purely about the internal I2S connection. With over 30 DAC boards, more than 8 USB boards / interfaces and AK4118 / AK4137 boards, I have never had any functional problems.

But we're talking about external I2S in connection with the industry standard LVDS (which is not tied to plugs or connections), the world's leading standard for external lossless cable transmission over short distances from Inter-IC applications. Incidentally, many devices at home also use LVDS (computers / laptops, monitors, televisions, data storage systems, etc.).
And there is the crux of the matter, no one has ever taken care of a standard, so the manufacturers have their own assignment.
That an HDMI cable is used for transmission is more of a coincidence, but above all it was available and cheap (plugs, sockets, cables). The transmitted signal has nothing to do with HDMI.

There are approx. 4 standard assignments via HDMI plugs that can be set for USB interfaces / sources (e.g. Topping U90, Singxer, Matrix, Gustard etc.).
Many keep crying that there is no standard.
If a manufacturer uses its own assignment of the HDMI connector, then he does not want his device to be compatible with other devices via this connection or to be used. That is the decision of this manufacturer, even if you can handle it with a special cable.

The transmission of I2S using LVDS is also superior to all other types of transmission (AES, SPDIF, optical), including USB (who came up with the idea of laying a very dirty power supply within the shielding of a 2-wire data line?).
So if you have a source device that outputs the digital audio data directly as I2S / LVDS and the cable path is less than 1-2 meters, then it is the technically best and simplest solution. E.g. with the internal tapping of the I2S signal in a CD player and connection via an I2S / LVDS HDMI converter.
Now comes the biggest BUT, there are currently only a few real source devices with an I2S interface.

However, if you do not notice any difference in sound between the various connections (this is also my experience, not always, but often), it does not matter in the end.
 

Roland68

Major Contributor
Joined
Jan 31, 2020
Messages
1,352
Likes
1,121
Location
Cologne, Germany
Nothing is being made better. At best you're getting a 1:1 reproduction, and at worst you're dealing with clock sync issue via jitter.

An external clock signal is added complexity and added risk, it is not beneficial unless your DAC which is receiving has a really ****** internal clock, but most have femto clocks when you get into "end game".

My smsl vmv d3 even has the connection for a 10mhz clock... but it's got 2 femto inside so zero point in using it.

This might be appealing with a topping dac, but ugh, lot of effort to get away from a crackle when playing non dsd after dsd file.
In your SMSL VMV D3 DAC you have 2 Accusilicon AS318B clocks (the same are also in my DAC), which is now a common high-quality standard.
A good 10MHz clock with a correspondingly high power consumption and a good power supply is 50% lower with a deviation.
You can also swap the internal crystals for the new Accusilicon AS338, whose accuracy, phase noise and short-term stability have been significantly improved.
But can you still hear that?
 

MacCali

Major Contributor
Joined
Dec 21, 2020
Messages
1,121
Likes
535
Nothing is being made better. At best you're getting a 1:1 reproduction, and at worst you're dealing with clock sync issue via jitter.

An external clock signal is added complexity and added risk, it is not beneficial unless your DAC which is receiving has a really ****** internal clock, but most have femto clocks when you get into "end game".

My smsl vmv d3 even has the connection for a 10mhz clock... but it's got 2 femto inside so zero point in using it.

This might be appealing with a topping dac, but ugh, lot of effort to get away from a crackle when playing non dsd after dsd file.
Crackle? We will find out I guess. My Pi2AES is almost setup. I got an i2s cable with an exact feed layout as the AES and equal input layout on my D70S.

So there should be no issues. Nor did I go out there and drop 300 on a “legit” i2s cable. Gustard has one for 32 dollars that I purchased, still over priced for a hdmi cable but whatever. At least everything matches up, which we hope to create the least issues.

Waiting for my AES cable to arrive, probably going to be another 2 weeks until it arrives. I’m sure I will see an improvement, but I will not tie it to the i2s rather the Pi2AES. My first streamer.
 
Top Bottom