• 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!

Study: Is I²S interface better for DACs than S/PDIF or USB?

Kal Rubinson

Master Contributor
Industry Insider
Forum Donor
Joined
Mar 23, 2016
Messages
5,273
Likes
9,789
Location
NYC
Once again: I2S is not a device to device connection standard.
Yup. When it was first used for external audio connections, the connectors used were 5-pin DIN and DB-25 (for I2Se).
i2se.JPG
 

sarumbear

Master Contributor
Forum Donor
Joined
Aug 15, 2020
Messages
7,604
Likes
7,314
Location
UK
I don't understand the question. The clock (SCK) defines when the data (SD) is valid. The word select (WS) is needed for the receiver to know where the sample boundaries are. If either one of these three signals is removed, what remains is useless. I don't know that anyone is ignoring anything.
I have not read about WS so far but been reading separate clock. Maybe I missed. However, as there is no written standard that I can find so far I asked if WS is ignored. If you know of any document that defines/explains how I2S as it is used externally by some manufacturers please post. Otherwise everyone is in the dark, I expect you as well.
 

mansr

Major Contributor
Joined
Oct 5, 2018
Messages
4,685
Likes
10,700
Location
Hampshire
I have not read about WS so far but been reading separate clock. Maybe I missed. However, as there is no written standard that I can find so far I asked if WS is ignored. If you know of any document that defines/explains how I2S as it is used externally by some manufacturers please post. Otherwise everyone is in the dark, I expect you as well.
Some of the manufacturers publish the pin assignments they use. Here's an example from Audio-GD:
hdmi-i2s.png


I assume SCLK here is the same as SCK in the I2S spec, SDATA is SD, and LRCLK is WS. MCLK would be the high-rate clock I mentioned earlier as being required by many DAC chips.
 

sarumbear

Master Contributor
Forum Donor
Joined
Aug 15, 2020
Messages
7,604
Likes
7,314
Location
UK
Some of the manufacturers publish the pin assignments they use. Here's an example from Audio-GD:
View attachment 155909

I assume SCLK here is the same as SCK in the I2S spec, SDATA is SD, and LRCLK is WS. MCLK would be the high-rate clock I mentioned earlier as being required by many DAC chips.
It is a shame that they haven't got together and created a standard. This is not the right way to treat a consumer!
 

Rottmannash

Major Contributor
Forum Donor
Joined
Nov 11, 2020
Messages
2,969
Likes
2,606
Location
Nashville
Some of the manufacturers publish the pin assignments they use. Here's an example from Audio-GD:
View attachment 155909

I assume SCLK here is the same as SCK in the I2S spec, SDATA is SD, and LRCLK is WS. MCLK would be the high-rate clock I mentioned earlier as being required by many DAC chips.
The manual for my D70 has a diagram of pin assignments.
 

mansr

Major Contributor
Joined
Oct 5, 2018
Messages
4,685
Likes
10,700
Location
Hampshire
The manual for my D70 has a diagram of pin assignments.
It does indeed:
1632827447112.png


Compared with the figure from Audio-GD, this has the opposite polarity on two of the differential pairs (red labels in the figure). It also has some additional signals. For extra fun, there is apparently a setting to switch the polarity of those two pairs.
 

Zog

Active Member
Joined
Mar 13, 2019
Messages
255
Likes
290
Maybe I'm completely missing the point and totally stupid but...

The XMOS USB interface in a USB desktop DAC converts USB > I2S that get send to the DAC chip.
So what is the benefit/added value of adding an extra box that you connect USB from the computer and outputs I2S through HDMI cable to the DAC?
I2S supports PCM 1536 and DSD 1024, is that really such a big deal compared to PCM 786 and DSD 512 that most XMOS based DACs support?
Using the Pink Faun 12S Bridge, as it is called, is an add-on card for a PC. Think of it as something like the Sound cards from Soundblaster or Asus. First it has the advantages of not converting to USB and the back to I2S in the DAC. Though to be fair these digital to digital conversions are all benign. Second, my understanding is that the signal is balanced from the get go.
 

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,632
Likes
10,205
Location
North-East
It does indeed:
View attachment 156048

Compared with the figure from Audio-GD, this has the opposite polarity on two of the differential pairs (red labels in the figure). It also has some additional signals. For extra fun, there is apparently a setting to switch the polarity of those two pairs.

I have two DACs supporting I2S over LVDS on HDMI. Amusingly, the left and right channels are swapped between them, so I have to fiddle with dip switches every time I swap DACs or flip analog interconnects between channels. Standardization? We don't need no freakin' standardization!! :)
 

DWI

Senior Member
Joined
Feb 19, 2021
Messages
495
Likes
437
My understanding is that and external SACD transport / DAC requires a proprietary connection to comply with Sony's rules for the format. So PS Audio's propietary I2S format is required for that compliance.

Of course they will say it sounds better, any manufacturer with any sense would make such a claim, it's not difficult to compare a PCM stream with I2S and usb or whatever and form your own opinion. You cannot compare with DSD as you can only use the I2S connection.
 

Lupin

Addicted to Fun and Learning
Joined
Feb 11, 2021
Messages
575
Likes
961
Using the Pink Faun 12S Bridge, as it is called, is an add-on card for a PC. Think of it as something like the Sound cards from Soundblaster or Asus. First it has the advantages of not converting to USB and the back to I2S in the DAC. Though to be fair these digital to digital conversions are all benign. Second, my understanding is that the signal is balanced from the get go.
Looked at the product page of the Pink Faun 12S Bridge.
Personally I don't see anything that would justify the €325 price tag for the basic version, with upgraded clocks the price goes even up to an incredible €1575 o_O
I have the feeling that the XMOS interface in my DAC is feeding the DAC chip an I2S signal that is just fine, especially since there's some extra conditioning before it reaches the chip. I'm sorry but I don't see how/why this Pink Faun would improve my setup.

Excuse my ignorance but... what exactly is a "balanced digital signal"?
And how/why will this make the Pink Faun worthwhile?
 

Roland68

Major Contributor
Joined
Jan 31, 2020
Messages
1,363
Likes
1,127
Location
Cologne, Germany
Looked at the product page of the Pink Faun 12S Bridge.
Personally I don't see anything that would justify the €325 price tag for the basic version, with upgraded clocks the price goes even up to an incredible €1575 o_O
I have the feeling that the XMOS interface in my DAC is feeding the DAC chip an I2S signal that is just fine, especially since there's some extra conditioning before it reaches the chip. I'm sorry but I don't see how/why this Pink Faun would improve my setup.

Excuse my ignorance but... what exactly is a "balanced digital signal"?
And how/why will this make the Pink Faun worthwhile?
Unfortunately, the high price is a fundamental problem and difficult to understand for end customers these days.
A card like the Pink Faun I2S Bridge sells maybe 100 pieces per year (or less). This means that the entire development costs must be allocated to this number of units. In addition, there are the low production numbers, which make the whole thing even more expensive. I estimate that the development costs will be over € 10-20,000, then you haven't produced anything yet. It looks different, of course, if you, like Asus, produce 100,000 cards, then such a card would cost well under € 100.
But in general, the last 2% cost more than the first 98%.

I would also be interested in whether it makes a difference, but currently I don't have the time to set up a PC for it.

With symmetrical transmission (regardless of whether analog or digital) the signals are mirrored (+/-) transmitted in parallel. This allows interfering signals or interference to be calculated out again. LVDS is also based on this.
 

Zog

Active Member
Joined
Mar 13, 2019
Messages
255
Likes
290
Unfortunately, the high price is a fundamental problem and difficult to understand for end customers these days.
High price to whom? To all end customers? What is a high price for one person is a cheap price for someone else.
 

Roland68

Major Contributor
Joined
Jan 31, 2020
Messages
1,363
Likes
1,127
Location
Cologne, Germany
High price to whom? To all end customers? What is a high price for one person is a cheap price for someone else.
Yes of course. It's always a matter of your attitude towards something, your priorities and your own possibilities.
And, of course, whether you are willing to pay the price for the last 2%.
 

BDWoody

Chief Cat Herder
Moderator
Forum Donor
Joined
Jan 9, 2019
Messages
6,948
Likes
22,625
Location
Mid-Atlantic, USA. (Maryland)

Killingbeans

Major Contributor
Joined
Oct 23, 2018
Messages
4,089
Likes
7,547
Location
Bjerringbro, Denmark.
Excuse my ignorance but... what exactly is a "balanced digital signal"?

It gets "split" into two differential signal to get high common-mode rejection? Can't really imagine it makes any difference (no pun intended).

The nice thing about AES is that it lets you have longer cable runs. It does not remove a mythical layer of awfulness :D
 

Lupin

Addicted to Fun and Learning
Joined
Feb 11, 2021
Messages
575
Likes
961
whether you are willing to pay the price for the last 2%.
So far I have not read anything that shows that it does anything meaningful.
In that regard the product is expensive no matter the price as it doesn't do anything.
 

Roland68

Major Contributor
Joined
Jan 31, 2020
Messages
1,363
Likes
1,127
Location
Cologne, Germany
It gets "split" into two differential signal to get high common-mode rejection? Can't really imagine it makes any difference (no pun intended).

The nice thing about AES is that it lets you have longer cable runs. It does not remove a mythical layer of awfulness :D
It is not about myths or beliefs.
This technology is all about being able to remove interference / interfering signals that scatter into the cable along the cable path from the useful signal. It doesn't matter whether the signal is digital or analog.
It actually has nothing to do with the devices, but with the external conditions. Ideally, it will make no difference whether you transmit the signal via AES / LVDS or, for example, via a coax cable. But in suboptimal conditions (sources of interference) and / or long cable runs, it makes a difference.

During these discussions, nobody should forget that a cable, regardless of whether LVDS (I2S), AES, Coax, USB, analog or digital, can transmit the data to a maximum of 100% (or an approximation to it), 101% is not possible. If it transmits less it is bad, if something is added it is wrong.
 

Roland68

Major Contributor
Joined
Jan 31, 2020
Messages
1,363
Likes
1,127
Location
Cologne, Germany
Ich würde das näher an die letzten 0% legen
Hope it was clear that this was only meant symbolically. It can also be 0.01%.
But...
Bisher habe ich nichts gelesen, was zeigt, dass es etwas Sinnvolles tut.
In dieser Hinsicht ist das Produkt unabhängig vom Preis teuer, da es nichts tut.
... it could also be worse than all other solutions. I don't know as I haven't tested it. As written in the previous post, whether or not it has an advantage can also depend on external circumstances.
But I don't allow myself to judge it either.

The title of this thread is: "Is I²S interface better for DACs than S / PDIF or USB?"
Although I have already tried a few things myself in this area, I cannot answer the question.
For me personally, and that is my experience from IT and industry, it is best to transmit a signal as directly as possible, without conversion and without detours on the shortest path.
If a source device can output I2S (over LVDS) directly and without internal detours, then the ideal would be because a DAC with this input could process the signal directly.
But if I achieve the same (or better) transfer with one of the other data transfer options AES, SPDIF (Coax / Optical) or USB, then it doesn't matter.
Depending on the source device, and of course also on the concept of the DAC, I2S can of course also be worse

In response to the thread title I would say it depends on various factors:
- Quality of the output (including conversion) and interface on the source device (what is missing or wrong here in terms of data ...)
- External influences (interference)
- Cable quality (purely physical / electrical)
- Quality of the transmission (also conversion) from the interface input to the DAC chip

It can depend on external factors as well as devices, transmission type and cables whether transmission with SPDIF (Coax / Optical), USB, AES or I2S is the better choice or solution.
In theory, and I also fear in practice, a simple and cheap transfer via USB or SPDIF (Coax / Optical) may or may not beat an expensive solution with I2S.
 
Last edited:
Top Bottom