What is the best way to get I2S output from streaming?

I am thinking about switching DAC to a R2R with I2S input, and then I would like to use I2S instead of USB. The DAC I am thinking about is Musician Pegasus which has I2S as a HDMI (LVDS) input. I am also using HQPlayer so the solution has to support NAA.

I currently use a tweaked/upgraded microRendu behind an opticalModule (fiber in), and run NAA on the microRendu. An easy way to solve I2S with streaming would be to just buy one of the USB → I2S converters out there like Matrix SPDIF-2 USB.
https://www.audiophonics.fr/en/digital-interfaces/matrix-x-spdif-2-usb-interface-xmos-u208-32bit-768khz-coaxial-aesebu-i2s-hdmi-lvds-p-11975.html

Another option would be to use some quality HATs with an RPi, like the reclocking module with the I2S transport module from Ian Canada:
https://www.audiophonics.fr/en/dac-and-interfaces-for-raspberry-pi/ian-canada-fifopi-q3-ultimate-fifo-reclocker-module-pcm-32bit-768khz-dsd1024-dop-p-14758.html
https://www.audiophonics.fr/en/dac-and-interfaces-for-raspberry-pi/ian-canada-hdmipi-transmitter-i2s-dsd-dop-to-hdmi-p-13982.html

And then upgrade clocks to something good like the ACCUSILICON AS318-B. I use LiFePo4 cells as power (pure cells), 20Ah, so I have power source covered in either case.

But which solution will sound best? In principle I like the RPi solution best since it totally gets away from USB, but actual sound quality is of course more important than preferences like that.

:dizzy_face:

Bored of ADI-2? :grinning:

I use the RME with class D amps (NC500) and slightly forward speakers, so sound is a little to analytical and forward, especially in combination with opticalModule and fiber which I use. A quality R2R DAC seems like a good step for me.

Since you mentioned HQPlayer, I’ll tag @jussi_laako to see if he will try to turn you away from I2S over cable… I think I know what he’ll say.

Holo Audio Spring2 is one of his favourites for use with HQPlayer. Features a special I2S input too if you really really want to use that.

image

While waiting for others to reply here you could also ask the manufacturers of these DACs for recommended I2S output steamers

1 Like

I know people using this. They recommend it.

Don’t think it would work with Spring2 though (not LVDS).

1 Like

And why do you specifically would like to have an I2S input and not choose for a R2R DAC with USB input behing a Raspberry Pi (like the Sonnet Morpheus for example)?

I have heard that I2S sounds better, and it’s even mentioned in one of the review for the Pegasus DAC. USB requires a little computer to handle (XMOS etc) which is bound to produce internal noise, I2S on the other hand is very lightweight.

1 Like

Since you already have Sonore products, why not just add an ultraDigital? I use one between my SOtM SMS-200 Ultra and a PS Audio DS DAC. It works great and is the preferred input for the DAC. The one thing to be careful of is that there is no common standard for the i2S pinout, so you need to make sure that whatever you use for the source i2S is compatible with you DAC.

1 Like

So far I’ve just seen measurements that show I2S increasing noise in the DAC output… Especially because external I2S sources have the clock at the wrong end of the connection (at the source, while it should be at the DAC side).

Based on what I’ve measured from Holo Spring USB inputs, there’s nothing to worry.

3 Likes

I have a friend who switched from USB to I2S on his Audio-GD master DAC and he said it was a quite big improvement, and then its also mentioned in one of the reviews of the Pegasus DAC that I2S sounds better. And then its always fun to play around with new tech :slight_smile:

@jussi_laako will the NAA image for Raspberry work with I2S HATs like the one from Ian Canada? Se first post for links.

Unless you have a DAC that ignores the clock coming in over I2S and uses its own…

1 Like

The I2S protocol has timing information, so I would assume that it should be clocked at the source and then (using the DACs clock) sent at the correct time. It should not be reclocked.

It depends on the DAC. The USB receiver on the original Holo Spring and on the Metrum Onyx was audibly inferior to I2S, I owned both. However, Holo upgraded their USB receiver in later production runs of the Spring and on the Spring 2, even though they still recommend I2S as their best source. I currently own a Sonnet Morpheus (on I2S) and elsewhere a Spring 2 (temporarily on S/PDIF but eventually on I2S when I rearrange the source), but I’ve not had time yet to seriously evaluate USB vs other inputs for either.

If a USB to I2S is used externally, like for example a Sinxger or Matrix SPDIF2 USB, then I would guess you just replace whatever is internally in the DAC from its USB input. That’s one of the reason I would prefer a more pure approach without any USB involved, like the stuff from Ian Canada.

On the other hand, RPi is not exactly quiet and even with separate power sources, reclocking board and isolator board its probably not 100% optimal.

There’s no reason these days, except design compromises, for a DAC USB receiver to be worse than external I2S; in fact, as @jussi_laako wrote, it should be better, at least with respect to jitter. When we are talking of multi-$K DACs, this should not be an issue, although unfortunately it still seems to be in some cases.

I think the heaviness of USB can make a difference. It needs a little CPU with memory to pack and pack up (XMOS or similar), which probably will affect oscillator and other electronics nearby. Having the USB handling away, like for example when using a USB → I2S converter, could be beneficial, but best (in my opinion) is to get rid of USB altogether.

Lets take an example: I upgraded the USB clock to a 24Mhz Crystek CCHD in my streamer. Naturally the clock in the DAC is used to clock the actual audio, but it still made a difference (and a pretty large one at that). I can’t explain it, but it shows that USB is far from perfect for audio. Maybe the better clock put less strain on the receiver, which leads to less electronic noise?

Then you have other problems, like how to deal with clock drift between two oscillators?

That’s the problem, that DAC needs to use the clock sent by the source… While clock should be at the DAC side.

Do you have some Jtest measurement data that shows this?

I’m curious to see all possible J-test measurement data between built-in USB and I2S for Holo Audio DACs.

While my Spring 1 is not quite as good as Spring 2, it is still very good:

vs

Yes, and it’s I2S interface is limited to 192k PCM sampling rate.

I’d first love to see the problem… :wink:

1 Like

Unfortunately I don’t have measurements, but some of those comparisons (it was several years ago so I don’t have everything fresh on my mind) were blind, with a subject who did not know which system they were listening to. Holo Audio updated the USB receiver on the Spring 1 at some point, do you know whether you have the updated one?

I’m not sure, but I think I have the older one… IIRC, it identifies itself as “xCore Audio” or similar.

While Spring 2 identifies as:
“Holo Audio UAC2.0 Gen2 Standard”

You can’t see it, you can hear it though :slight_smile:

Will the NAA Image works on an RPi with I2S output like with the Ian Canada boards?