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

[JSmith]

Uh-huh.

Looks like some members are forgetting about a few things called data transfer and error correction...

https://en.wikipedia.org/wiki/Error_detection_and_correction



JSmith

 

[mansr]

Looks like some members are forgetting about a few things called data transfer and error correction...

https://en.wikipedia.org/wiki/Error_detection_and_correction

It all comes back to our friend Claude Shannon, as always.

 

[BDWoody]

is there something you specifically don't comprehend?

On that subject, I'm going to put my $$ on @mansr in terms of who comprehends what...

 

[JSmith]

it's time for them to understand

JSmith

 

[chris719]

Looks like some members are forgetting about a few things called data transfer and error correction...

https://en.wikipedia.org/wiki/Error_detection_and_correction



JSmith

Actually, USB isochronous transfers have no error correction. That doesn’t mean that anything pedrob said is correct, of course .

 

[Zog]

What is an I2S cable? There is no such official thing. I2S is an internal bus to sane people. If you mean an HDMI cable carrying LVDS pairs of the I2S signals, then fine... but it's an HDMI cable.

You define by structure. But is equally valid to define a thing by its function. Also, I find the put-down offensive. Just 'I2S is an internal bus' is sufficient.

 

[chris719]

You define by structure. But is equally valid to define a thing by its function. Also, I find the put-down offensive. Just 'I2S is an internal bus' is sufficient.

There is no such thing as an I2S cable. When you find a real standard that mentions one, please let me know. I have a copy of the I2S standard and it doesn’t cover cables because it is a point to point IC to IC bus.

 

[k9gardner]

The thing is, even if it's all aural chimera, you probably want to spend dollars wisely. Support good engineering at a particular price point. What gets me are companies selling overpriced junk that offers no value at all, against lesser priced competition. This doesn't mean that high priced gear has no reason for existence. There are a lot of reasons one might want to buy an expensive solid state amp: quality of construction, dealer and manufacturer support, pride of ownership, visual aesthetics (backlit blue meters)... etc.

Loudspeakers are really the only modern day component where anyone can reliably tell differences. Loudspeakers are really the last remaining subjective component. Coupled to your listening room, which could make the same loudspeaker sound different. Of course, the recording itself makes a huge difference, but that seems to be a given. Possibly, just possibly, amplifiers, if the device is tweaky to begin with, and horribly mismatched to speakers--say, an SET amp connected to a crazy low impedance speaker, one that requires high current availability. Then, for old timers like me, phono carts (but probably not phono stages) and turntable/arm combos could make a difference. Cassettes and open reel decks. But who cares about them? Anything mechanical, producing lots of distortion. But then we are back to loudspeakers.

This I think is the first post in this thread that mentions cost. Dollars. For most mortals, cost matters, and we want to get the best bang for our buck. I am very green in the current subject area, please bear this in mind, but I am a systems guy and I am working through discussions like this to pick up as much as I can.

My goal is to understand these digital audio problems and solutions well enough to not make obvious mistakes. Right now, in broad strokes, I have a CD player, plugged into an amp, which drives my speakers. I would like to transition to a media server, containing all my CD musical content, and to be able to send it to the same amplifier and speakers, in the most robust way possible (for the price). That's really it. (Ok, ideally I'd like further to be able to route that content to the bedroom, or the office, instead of the living room, but I understand that that's a whole 'nother discussion.)

With that goal in mind (assuming that that's the goal of others here as well...), isn't it true that connecting a DAC to my media server via USB means that whatever else the USB bus is dealing with – a mouse, a webcam, a printer – would potentially compromise USB bandwidth and hence the signal that I want as clean as possible for my DAC? Whereas I2S is dedicated to one thing, and so would be unperturbed by all of those other activities? Shouldn't we be doing an A-B comparison between USB and I2S in a worst-case scenario test, rather than best-case?

Second (remember I'm coming from a cost-aware perspective), if I'm looking specifically at various R-pi HAT configurations, wouldn't a DAC HAT, sitting on the R-pi I2S, outputting directly to the amplifier, be the shortest path? Why get into further translations, S/PDIF, TOSLINK, any of that, if not necessary?

I think these questions are in line with the OP, but please don't hesitate to point me toward a different thread or forum if they seem beside the point.

 

[Zog]

There is no such thing as an I2S cable. When you find a real standard that mentions one, please let me know. I have a copy of the I2S standard and it doesn’t cover cables because it is a point to point IC to IC bus.

Oh dear. I am not sure you have grasped the difference between assertion and argument. I mention function and you respond with assertion. Or should I play the game? Me - yes it is I2S, You - no it is not, Me - yes it is, You - no it is not. That would be infantile, right? Also we are not talking about IC to IC - so go away with your strawman. We are talking about a connection between a source and a DAC, not within a CD for example.

If you want to debate my first comment then tell me why it is not legitimate to define a term by function rather than structure. (An interesting concept) - so please chip in if you are not just another 'wanna win' person.

 

[chris719]

Oh dear. I am not sure you have grasped the difference between assertion and argument. I mention function and you respond with assertion. Or should I play the game? Me - yes it is I2S, You - no it is not, Me - yes it is, You - no it is not. That would be infantile, right? Also we are not talking about IC to IC - so go away with your strawman. We are talking about a connection between a source and a DAC, not within a CD for example.

If you want to debate my first comment then tell me why it is not legitimate to define a term by function rather than structure. (An interesting concept) - so please chip in if you are not just another 'wanna win' person.

There's no strawman. There is no such thing as an official I2S cable or any standard that brings I2S itself outside of a box. Your lack of understanding is not anyone else's problem. I'm not going to argue semantics. If you go to Best Buy or Amazon it's called an HDMI cable.

 

[Zog]

There's no strawman. There is no such thing as an official I2S cable or any standard that brings I2S itself outside of a box. Your lack of understanding is not anyone else's problem. I'm not going to argue semantics. If you go to Best Buy or Amazon it's called an HDMI cable.

Oh come on; stop this talking down Best Buy Amazon nonsense - No, just no ... and surely you know better. That is a distraction. You ought to know better than this. You do know this august forum does not take Amazon as an authority!

Your hint ..."I'm not going to debate..." to stop debating is not accepted. Let's debate your original concept - remember that? The definition of I2S cable.

 

[chris719]

Oh come on; stop this talking down Best Buy Amazon nonsense - No, just no ... and surely you know better. That is a distraction. You ought to know better than this. You do know this august forum does not take Amazon as an authority!

Your hint ..."I'm not going to debate..." to stop debating is not accepted. Let's debate your original concept - remember that? The definition of I2S cable.

I'm not sure why you think this is up for debate or that I want to debate with you on something you don't have any background in. Have a good evening.

 

[JSmith]

wouldn't a DAC HAT, sitting on the R-pi I2S, outputting directly to the amplifier, be the shortest path?

Yes... but DAC hats simply don't perform as well as a well measuring external DAC, which is why many use a Pi -> DAC.



JSmith

 

[mansr]

With that goal in mind (assuming that that's the goal of others here as well...), isn't it true that connecting a DAC to my media server via USB means that whatever else the USB bus is dealing with – a mouse, a webcam, a printer – would potentially compromise USB bandwidth and hence the signal that I want as clean as possible for my DAC?

When a USB device, such as a DAC, that uses isochronous transfers is opened by a client application, the necessary bandwidth is reserved in order to prevent the situation you envision. Multiple isochronous devices can each reserve a portion of the total available bandwidth. Whatever remains is used for lower priority bulk transfers to/from storage devices, printers, etc.

 

[somebodyelse]

Second (remember I'm coming from a cost-aware perspective), if I'm looking specifically at various R-pi HAT configurations, wouldn't a DAC HAT, sitting on the R-pi I2S, outputting directly to the amplifier, be the shortest path? Why get into further translations, S/PDIF, TOSLINK, any of that, if not necessary?

Given that the transitions are demonstrably lossless (assuming a decently engineered DAC with nothing broken) what makes you think these transitions are any more of a problem than the several that go on in the Pi between raw storage, through the processor and its memory, and on to whatever interface you use? There are specific reasons you might pick one interface over another, but getting the data to the destination correctly probably isn't one of them. From a cost aware perspective have a look at the relative prices of similarly performing DACs with each type of interface. The HAT style DACs tend to cost more for a given level of performance - presumably for reasons of market economics and production volume. Even so, their performance may be good enough, and non-audio considerations may make them a better solution than an external DAC or dongle.

 

[Migel83]

The topic of I2S in general is a bit thin and even if it goes a complex way to be jitter free you can still do a lot wrong.
Too long a cable has already been mentioned several times in practice.
Especially with the clock, this can cause disturbances.
Therefore, it should often be as short as possible.

As has already been mentioned, it has not yet been decided.

With the Raspberry, you always hear about the bad power supply, which affects the technology.
So it's better to go straight for the Singxer or Matrix Audio.

The third thing is probably the compatibility technology, which also has an influence.
There are also statements that Gustard runs better with product X than with Y.

The matter is not new but still unexplored.
It could still take a few years until there is agreement or not.

Apart from the measurements, I would rather advise you to test with the Hdmi cable in the length that you personally like.
And the last advice is to get help from people who know about it rather than making false claims.

I will pair the Su 2 with the SDA2 C both from Singxer and see what happens.

 

[EdW]

Yes... but DAC hats simply don't perform as well as a well measuring external DAC, which is why many use a Pi -> DAC.



JSmith

HAT DACs can be very low cost. If performance is less important than cost this could be the deciding factor. But I too would always recommend spending a little more to get a USB DAC which will offer substantially more performance and are generally manufactured in very high volumes for good value for money. High end HAT DACs, comparable in price to a good low cost USB DAC underperform relative to the USB DAC.

Also to consider:

1. USB DAC is shielded from digital noise from the Raspberry Pi
2. Good USB DACs can be powered by commodity SMPS adapters or have a power supply built in. No nonsense about the need for over the top linear power supplies for maximum quality
3. With most USB DACs you can input TOSLINK or SPDIF coax as well, allowing you to connect in your TV and CD player. This is a very significant advantage

Disadvantage: you need to buy a RPI4 rather than a RPI3 for guaranteed satisfactory operation over USB. So marginally more expensive

 

[Zog]

I'm not sure why you think this is up for debate or that I want to debate with you on something you don't have any background in. Have a good evening.

Well done, another slur and another strawman. You do not know my knowledge in DACs and so stop the gratuitous insults dude. It was your put-down to a third party that lead me to call you out. The 'debate' is a concept you should know better: think argument (in the sense of applying reason) -vs- assertion.

Go away - in fact I am putting you (and only you) on 'Ignore' - so reply to the group if you must, but I will not see it.

 

[chris719]

Well done, another slur and another strawman. You do not know my knowledge in DACs and so stop the gratuitous insults dude. It was your put-down to a third party that lead me to call you out. The 'debate' is a concept you should know better: think argument (in the sense of applying reason) -vs- assertion.

Go away - in fact I am putting you (and only you) on 'Ignore' - so reply to the group if you must, but I will not see it.

I read your posts in this thread, thus I can firmly conclude that you have no idea what you are talking about. I'm not trying to insult you, you just don't.

 

[sumpao]

I read your posts in this thread, thus I can firmly conclude that you have no idea what you are talking about. I'm not trying to insult you, you just don't.

I read two of you post not still don't understand what the issue