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

[mansr]

Latency can play a major role here.

Sub-millisecond differences in latency are inconsequential in audio applications.

 

[Markus 1204]

Hello, yes, I use Gigabit and CAT 6. It works fine with HifiBerry or Yamaha, only with Minidsp LAN or doesnt work. Latency could be possible. If I do a test sweep from HifiBerry via Spdif to the Minidsp, the signal is at an other position in the graph. It's faster as with LAN. But why? Interesting is, that with all other inputs, the signal is perfect and really good with using Dirac. With the LAN Input, the signal without DIRAC is better. But not so good like signals from the other inputs. I think something between LAN Input and DAC is not good. If it's hardware or software, I don't know.

 

[Markus 1204]

But there is something more wrong and I'm in contact with the support....but I can see in the Forum of MiniDsp, the problem is old and unsolved..... the channels are sometimes switching the sides. Left signal is coming from the right and right from the left. At the moment, I'm really not impressed.

 

[Zog]

Sub-millisecond differences in latency are inconsequential in audio applications.

A one millisecond frequency is an octave below a two millisecond frequency. Nicht Wahr? Or is this the wrong context?

 

[somebodyelse]

A one millisecond frequency is an octave below a two millisecond frequency. Nicht Wahr? Or is this the wrong context?

Wrong context I think - this is latency in network packet arrival, and unless MiniDSP have done something very unusual with the MPD configuration, its buffers will be much bigger than required to cover that.

 

[mansr]

Wrong context I think - this is latency in network packet arrival, and unless MiniDSP have done something very unusual with the MPD configuration, its buffers will be much bigger than required to cover that.

Right. The time required to transmit a maximum-size (1500 bytes) Ethernet frame at 100 Mbps is 0.12 ms. That's not enough to matter for audio streaming.

 

[Zog]

Wrong context I think - this is latency in network packet arrival, and unless MiniDSP have done something very unusual with the MPD configuration, its buffers will be much bigger than required to cover that.

Thanks for that. Trouble is: can you bring this down to my level. 'network packet' and MiniDSP mean nothing to me.

 

[somebodyelse]

Thanks for that. Trouble is: can you bring this down to my level. 'network packet' and MiniDSP mean nothing to me.

I'll have a go. If you follow the quotes back up the thread you'll see mansr and Gradius were discussing an issue in the network streaming part of the MiniDSP SHD DAC and streamer. Networks transfer small chunks of data known as packets, and some variation in when they arrive is to be expected as that's how networks work. Streaming audio requires the data to be split into these packets, sent, then reassembled by the receiving end, which keeps a buffer to cover the maximum expected delay in the network. As I understand it the receiving application used by MiniDSP in the SHD is called MPD, and by default keeps a buffer big enough for several seconds of audio. Packets would have to be delayed by long enough for the data in the buffer to run out before the DAC would see any difference, so sub-millisecond changes in packet arrival are invisible to the DAC.

 

[pedrob]

Let me be clear: I think PS Audio products are shown to be poor, and Paul McGowan appears to be a bullshitter.

Those letter segments are the pits. Certainly doesn't provide confidence in the business. I don't own any of its products and I don't think I ever will.

 

[sumpao]

I using Gustard U16 feed I2S to my Audio GD R 28

It sound different. I like it better than USB direct to dac.

Mostly I don't here any different in digital chain but this does

 

[pedrob]

I am certain the I²S cable has better shielding than the USB cable, particularly as those with high throughput require a superior level of shielding to achieve their rating.

Before all the sceptics start quoting digital is just 0's and 1's, it's time for them to understand there is very little difference between the way analog and digital signals travel through cables. In fact the main difference is analog is sine wave, while digital is square wave. If you accept that shielding is necessary so that sine waves are not contaminated, then there is no reason to suggest square waves are any different. If EMI and RFI did not exist even at miniscule levels, there would be no need for shielding.

 

[mansr]

Uh-huh.

 

[pedrob]

Uh-huh.

is there something you specifically don't comprehend?

 

[UKPI]

Before all the sceptics start quoting digital is just 0's and 1's, it's time for them to understand there is very little difference between the way analog and digital signals travel through cables. In fact the main difference is analog is sine wave, while digital is square wave. If you accept that shielding is necessary so that sine waves are not contaminated, then there is no reason to suggest square waves are any different. If EMI and RFI did not exist even at miniscule levels, there would be no need for shielding.

Electric signals at different frequencies behave differently. Cables that are appropriate for sending analog audio signals can be a suboptimal choice for sending digital signals because of that "sine wave vs square wave" characteristic. Read about Transmission line theory and skin effect if you would like to know more. https://www.intel.com/content/dam/www/programmable/us/en/pdfs/literature/wp/wp_sgnlntgry.pdf

Yes, proper engineering is necessary for digital signals to be just 0s and 1s. However, that part is already done by countless manufacturers. Practically noiseless and distortionless final analog output signals from various DACs using different interfaces are the proof.

 

[pedrob]

Perhaps I should have explained the square wave isn't audio but digital and in the respect it is a switch; on/off which represents 0's and 1's.

 

[pedrob]

I'm just talking about cables. Once the signal is contaminated nothing is smart enough to distinguish it and remove it.

 

[chris719]

Perhaps I should have explained the square wave isn't audio but digital and in the respect it is a switch; on/off which represents 0's and 1's.

Unfortunately I don't know how to put this more nicely than this: it's clear that you know so little on this topic that you aren't even aware of the limitations of what you know.

It's a deep topic, so there's no shame in that, but you've made a few statements here that aren't correct and you seem dug-in already...

 

[chris719]

I am certain the I²S cable has better shielding than the USB cable, particularly as those with high throughput require a superior level of shielding to achieve their rating.

Before all the sceptics start quoting digital is just 0's and 1's, it's time for them to understand there is very little difference between the way analog and digital signals travel through cables. In fact the main difference is analog is sine wave, while digital is square wave. If you accept that shielding is necessary so that sine waves are not contaminated, then there is no reason to suggest square waves are any different. If EMI and RFI did not exist even at miniscule levels, there would be no need for shielding.

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.

I2S is slow. USB 3.x SS 5Gbps and SS+ 10Gbps are real things, you know. As is Thunderbolt and DP Alt Mode over USB cables. HDMI cables might be theoretically better but the difference manifests itself at incredibly high rates, nothing that will impact the integrity of this janky I2S connection.

In any case, it has little to do with their shield construction...

 

[sumpao]

is there something you specifically don't comprehend?

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.

I2S is slow. USB 3.x SS 5Gbps and SS+ 10Gbps are real things, you know. As is Thunderbolt and DP Alt Mode over USB cables. HDMI cables might be theoretically better but the difference manifests itself at incredibly high rates, nothing that will impact the integrity of this janky I2S connection.

In any case, it has little to do with their shield construction...

I2S between device now using HDMI (PS AUDIO) or Eternet lan . I don't know much about cable and I don't think it make any different in the chain if the connection is fit properly and the cable is well build.

 

[sumpao]

There a D to D device like Gustard U16 . Denefrip some HERMES or Gaia or Audio GD DI

If your dac have I2S connection even you cannot measure it did anybody here try it and not hear a different?

I don't say it will sound better but would you not hear it?