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

[tom.chen]

I can't digital capture I^2S.

Hi Amirm
Did you configure the I2S sequence of SU1?
I2S has many sequence.
SU1 can work well with X26,but you need to set it in correct configuration.

 

[tom.chen]

Because the configuration is incorrect, LRCLK is reversed.
The data of left and right channels are delayed by one sampling period.
So there was a phase delay of 8.168 degrees.
8.168*44.1=360.2088

 

[pkane]

Because the configuration is incorrect, LRCLK is reversed.
The data of left and right channels are delayed by one sampling period.
So there was a phase delay of 8.168 degrees.
8.168*44.1=360.2088

That's correct. There's a set of dip switches on the bottom that must be set properly for the HDMI connector pin-out on the DAC side.

 

[tom.chen]

Why does the I2S interface will be very popular?
Because it can support very high rates.
PCM 1.536M and dsd1024 can work well too.
now many I2S interface using LVDS signal and HDMI interface.
as we know ,the LVDS can work in 400mbps or 600mbps.
the HDMI-I2S has 4 signal channels.(with mclk)
In theory, it can provide transmission capacity of more than 1.6Gbps.

 

[tom.chen]

sorry for my poor english.

 

[March Audio]

Only in implementation. The RPi has I2S wired straight from the SoC to the GPIO header. If you’re going to make a DAC that’s a Pi hat, I can’t think of much reason to not use it, as it’s right there and ready to use.

My recollection is that the quality of the the RPi i2s clock is total pants and cant be integer divided into 44.1kHz.

 

[Dant]

My recollection is that the quality of the the RPi i2s clock is total pants and cant be integer divided into 44.1kHz.

Ah, I didn't know that. Can't say I'm surprised though, RPi Foundation added it to the GPIO for sensor/device connections, not originally intended for audio.

 

[amirm]

Here are three sweeps of frequency versus interchannel phase differential at three sample rates:

So the differential is both frequency and sampling rate dependent.

 

[amirm]

Hi Amirm
Did you configure the I2S sequence of SU1?
I2S has many sequence.
SU1 can work well with X26,but you need to set it in correct configuration.

Hi there. You were right. Setting switch 6 to on fixed it. I will update the review.

Silly me, I thought once we got audio from the DAC, the configuration was correct. Not so.

 

[amirm]

Can we see the results with Gustard's GPLL off?

Edit: Sorry, I meant ESS ASRC/jitter correction off when the DAC is fed the data over I2S, ie the DAC runs in sync mode.

Here are the results with GPLL bypassed:

While not audible either way, jitter is worse with S/PDIF when you turn off GPLL.

 

[tom.chen]

Hi there. You were right. Setting switch 6 to on fixed it. I will update the review.

Silly me, I thought once we got audio from the DAC, the configuration was correct. Not so.

ok,thanks.
hope you have the correct test.

 

[la2ygoo]

Here are the results with GPLL bypassed:

View attachment 23913

While not audible either way, jitter is worse with S/PDIF when you turn off GPLL.

Hi Amirm
This dx3pro spdif provided by Singxer SU-1 or Gustard U12 or Apx555?

 

[Nango]

Why does the I2S interface will be very popular?
Because it can support very high rates.
PCM 1.536M and dsd1024 can work well too.
now many I2S interface using LVDS signal and HDMI interface.
as we know ,the LVDS can work in 400mbps or 600mbps.
the HDMI-I2S has 4 signal channels.(with mclk)
In theory, it can provide transmission capacity of more than 1.6Gbps.

Ok, by when the PCM 3.072 ??

 

[rajapruk]

If someone could tap out the raw I2S streams from a cheap cinema receiver to get the pure digital signal out of multichannel content, then yes, dac with I2S input could make sense.
There was a guy in Norway trying to do it on a entry level Marantz receiver. He never posted any success though.
Also Audiopraise has a chip extracting 7.1 I2S from OPPO UHD-bluray player.
Then HDMI-transfer can be avoided in the audio chain for multichannel content.

 

[anmpr1]

So Paul McGowan, please don't spread misinformation...

Forget it. Paul knows which side of the bread is buttered. His company's reason for existence is to hawk his own brand of news from the moronosphere. On the other hand, maybe you're doing it all wrong. Maybe it would 'measure better' if you had connected everything to his ten thousand dollar DirectStream Power Plant 20(™), which (it says here) "allows your most coveted components to unleash their full potential, no longer hindered by dynamic drops in AC caused by the unavoidable resistance of the miles of wiring between you and the power station."

 

[amirm]

Hi Amirm
This dx3pro spdif provided by Singxer SU-1 or Gustard U12 or Apx555?

Sorry that was a typo. Should say dac-x26 spdif from su-1.

 

[Veri]

Here are the results with GPLL bypassed:

View attachment 23913

While not audible either way, jitter is worse with S/PDIF when you turn off GPLL.

Hmm so objectively, the I2S does have the better jitter specs in this case?

 

[MZKM]

Hmm so objectively, the I2S does have the better jitter specs in this case?

If you turn off it’s jitter reducing feature, sure.

From what I could tell, the only negative to having it active (especially on High), is that it loses lock after X amount of jitter is present. Other than that, I see no reason as to want it off.

 

[gvl]

Is there an option on the X26 to disable the ASRC on the ESS chip so it will use MCLK from I2S? Would be interesting to compare I2S with ASRC on/off as this would be a real test between internal and external clocks.

 

[nm4711]

My recollection is that the quality of the the RPi i2s clock is total pants and cant be integer divided into 44.1kHz.

This is correct. But you can configure the RPi to be the clock slave. So you can use an external clock to avoid this problem.
Hifiberry uses the RPis i2s output this way since 2014. Atleast with their higher class products.