Just for fun: hypotetic ideal stereo digital audio streaming?

[robin79]

Disclaimer: I'm a newbie and not an expert in electronics at all, I just read a lot here and I'm a curious guy... so please, don't blast me if I write silly things

The game is to imagine an ideal-yet-minimal configuration (in electronics terms) for streaming digital stereo.
On the basis of my readings, this is my hypothetical chain:

Source
A network streamer with two IIS output ports via HDMI connectors.
Each port delivers only one channel of the stereo, thus: left ch. IIS port and right ch. IIS port.
The two ports share the same high quality master clock (IIS is in master mode).
Each port is connected to an active speaker via an HDMI cable.

Active full-range speakers (described as one)
The speaker offers an IIS input via HDMI connector.
Is full range (no need for sub), say a three or four way system (not relevant, but likely).
Each woofer/twitter of the speaker has a dedicated true-digital amplifier (like DirectDigital® from NAD or similar).
There is no actual cross-over, each true-digital amp receive its proper audio band via DSP processing and is optimized for its load and band.
The digital signal of each amp goes through an optional dedicated DRC (Digital Room Coorection) phase before the amplification step and the necessary smoothing to enter the analogue domain as late as possible.

How far is this from ideal in your opinion?
Who wanna play?

 

[NTK]

I2S was invented as a chip-to-chip communications interface standard, and is not intended as an external link connecting boxes via cables. There is no good reason to use it (and many not to use it) to transmit signals between two boxes, especially at the distance you are thinking about (streamer to speakers).

You may want to read this article by Bruno Putzeys on why "direct digital" isn't a good idea. Note that the recent top of the line NAD integrated/power amps use Hypex and Purifi class D amp modules, instead of "direct digital", and these class D amp modules are designed by Bruno.
https://www.audioholics.com/audio-amplifier/the-truth-about-digital-class-d-amplifiers

 

[robin79]

I2S was invented as a chip-to-chip communications interface standard, and is not intended as an external link connecting boxes via cables. There is no good reason to use it (and many not to use it) to transmit signals between two boxes, especially at the distance you are thinking about (streamer to speakers).

You may want to read this article by Bruno Putzeys on why "direct digital" isn't a good idea. Note that the recent top of the line NAD integrated/power amps use Hypex and Purifi class D amp modules, instead of "direct digital", and these class D amp modules are designed by Bruno.
https://www.audioholics.com/audio-amplifier/the-truth-about-digital-class-d-amplifiers

Thank you very much NTK for your feedback!

Starting from the end, I've read (two times) the article you shared, I can't get it completely because of my limited knowledge, what I have deducted is in terms of preconception and the complexity of the real world: "direct digital" is assumed to be superior on the preconception that the more the signal stays out of the analog domain the better it is, but implementation wise a digitally controlled class-D amplifier presents more inherent deficiencies than an analog controlled one, in a measure that makes the latter perform better.
Is this the bottom line, more or less?

To be honest, I am (was?...) one of those convinced that "less analog" implies "less distortion" and thus more quality: not having real notions I'm stick to abstract concepts
But actually I have too been observing the greatest companies moving away from the "direct digital" after an apparent initial enthusiasm (like NAD, but also Nuforce/Nuprime and others) or sticking to AB (like Benchmark Media Systems and Hegel) and I was wondering why: maybe Bruno's diagnosis is the reason, since anyway "they are largely very complicated analog design exercise" and the implementation counts more than the ideal approach.

Regarding the I2S topic, I have read a bit after your comment and I now understand my mistake: I have been influenced by the fact that many producers are offering I2S input ports on their DACs and I had also read that, through a clear separation of the clock signal from the audio data, the I2S protocol would enable the source component (the streamer, in my model) to act as a good master clock, in terms of jitter, to the slave components (the speakers).
Before "electing" I2S I had thought about SPDIF via asynch-USB, SPDIF via optical, SPDIF via coax and AES3 via XLR: I have filtered out those options for the following reasons

• having to feed two different DACs (one per speaker), I wanted those DACs to use the same master clock (asynch-USB clearly out, since it relies explicitly on the DACs' internal clock)
• I thought that an audio-data-only signal would be better (USB out again, because delivers power and audio data inside the same cable)
• all the considered options bring the clock info "sunk" in the audio data and I thought that this would be sub-optimal (so... all SPDIF-based solutions out)

I2S, with a separated clock signal and only audio data in the stream, seemed to me a good choice... I didn't know about the fragility of the signal and the consequent length limitations.

So, based on your contribution, I would apply these changes:

• two AES3 outputs, one per channel, via XLR (for lower noise on the signal)
• the same dedicated external master clock brought to both the speakers for re-clocking, ignoring the clock from the AES3 signal
• well-known class-D modules (like Purifi) instead of direct-digital

and all the rest unchanged.
Would it be "ideal" this way?

Thank you again for helping me to learn a bit, any further correction would be really appreciated.

PS: another member posted a reply quoting an existing product, a pair of high-end active monitos with built-in DAC, AES3 input and AES3 pass-through (I don't see his reply anymore... I don't understand why, was it removed because a brand was quoted?); I really don't get the "pass-through" solution: the audio signal reaches one of the speaker firstly, than runs though one more cable of unknown lenght and only then reaches the second speaker... how can this be desirable? How can the two DACs in the speakers work in synch with each other? And how can they receive the same signal with double of the cable lenght and the pass-through hardware in the middle? My ignorance again, I guess...