I2s vs USB

[coot]

Given a server such as Aurender N10 with I2s output in addition to usb and a DAC with inputs for both, is it better to go with direct i2s connections or hot-rodded usb specifically including additional devices such as Regen and/or power supplies, etc., etc.

[Blizzard]

Given a server such as Aurender N10 with I2s output in addition to usb and a DAC with inputs for both, is it better to go with direct i2s connections or hot-rodded usb specifically including additional devices such as Regen and/or power supplies, etc., etc.

 

Definitely I2S. No matter how much you polish a turd, it will never shine like a diamond.

[Blizzard]

But unfortunately the Aurender doesn't have a I2S output option. I'd consider the Sonore Signature Rendu. I haven't personally auditioned, but it would most likely smoke the Aurender if compared head to head.

[Norton]

I think of I2s mostly as an interface between upmarket transport/DAC pairings from the same manufacturer. Excuse the dumb question, but is it a universal standard, so that if I have an I2s input DAC, it will work with any I2s output source, presuming they are compatible as to audio format?

[Blizzard]

I think of I2s mostly as an interface between upmarket transport/DAC pairings from the same manufacturer. Excuse the dumb question, but is it a universal standard, so that if I have an I2s input DAC, it will work with any I2s output source, presuming they are compatible as to audio format?

 

PS audio tried making their I2S over HDMI a standard. Quite a few adopted it so far like Sonore, Wyred 2 sound etc.

 

Jesus from Sonore has a good list posted on another thread.

 

In order to take advantage of it, you must have both a DAC and a server/renderer/transport that has the connection. Some companies like MSB tech decided to use RJ45 with Ethernet cable instead.

 

IMO every decent DAC and Server/renderer/transport should have this option available. It future proofs DAC's because it allows direct connection to the DAC chip or internal FPGA. USB interfaces keep getting better every 6 months. With DAC's that have built in USB, 2-3 years down the road you are stuck with a DAC that still has awesome potential but is bottlenecked by obsolete technology. So then they offer a "supreme" or something version that's pretty close to the same thing but new USB interface. And if you want to upgrade your stuck with an old DAC nobody wants. If your lucky you can fetch half of what you paid on Audiogon.

 

With external I2S, if a hot new renderer comes out, your not bottlenecked by the obsolete USB interface. Just look at some of the DAC's from just 3-4 years ago with USB interfaces. Most DAC's were using the M2 highface. I remember all the raving about how awesome it was. A year later everyone's talking about how it's junk and the Xmos chips are better. Then the Amanero came along. Now Xmos has a new generation chip which I think is their 4th generation in the last 3-4 years.

 

Now all the rage is band aids like the REGEN that are designed to overcome the shortcomings that aren't present to begin with using I2S

 

When will the madness end?

 

 

 

At least some companies like MSBtech have their USB interfaces on modular plugin boards. IMO the only way it should be done.

[One and a half]

I2S is as far as I know a protocol that is found in the DAC only with really short distances like 2inches something.

 

I2S can be transported via LDVS for longer transmissions, say 1-2m from a server to a DAC. PSAudio & Sonore collaborated and use a HDMI cable, but why on earth use such a crappy connector which can fall out at any time.

 

Technically I2S/LVDS is far better than USB any day, but the world changes slowly...

[Superdad]

Technically I2S/LVDS is far better than USB any day, but the world changes slowly...

 

Just remember that with external I2S interfaces the DAC then gets slaved to the source's clock (MSB being the only folks I know of who recently added a clock-out to their I2S input). Contrast that with USB inputs where the DAC's master clock (just millimeters from the DAC chip) really is Master.

 

Besides, how are people getting output from their computers? USB, or if you embrace DLNA, Ethernet. (Though the Signature Rendu is a wonderful Ethernet>I2S rendering bridge and has the nice Crystek CCHD clocks.)

[Blizzard]

I2S is as far as I know a protocol that is found in the DAC only with really short distances like 2inches something.

 

I2S can be transported via LDVS for longer transmissions, say 1-2m from a server to a DAC. PSAudio & Sonore collaborated and use a HDMI cable, but why on earth use such a crappy connector which can fall out at any time.

 

Technically I2S/LVDS is far better than USB any day, but the world changes slowly...

 

Yes the world changes very slowly. MSBtech has been using I2S over LVDS since the 90's. And it's still their best way. Although they have made a few improvements over their original implementation.

 

USB for audio only came about as a way to get audio from a general purpose computer into a DAC.

[Blizzard]

Just remember that with external I2S interfaces the DAC then gets slaved to the source's clock (MSB being the only folks I know of who recently added a clock-out to their I2S input). Contrast that with USB inputs where the DAC's master clock (just millimeters from the DAC chip) really is Master.

 

Besides, how are people getting output from their computers? USB, or if you embrace DLNA, Ethernet. (Though the Signature Rendu is a wonderful Ethernet>I2S rendering bridge and has the nice Crystek CCHD clocks.)

 

Your wrong about the clock slaving. Most implementations don't even send the master clock from the source. The data gets clocked by the master clock when it gets into the DAC. It's a better way all around to do things. USB for audio is a compromised alternative designed to interface general purpose computers to DAC's. It's not a professional solution like I2S over LVDS.

 

The world is going to Ethernet renderers anyways. So won't be long before USB is phased out. This is a far better solution for those who want to use their general purpose computers as servers.

[Blizzard]

Here's Jesus's database of I2S over LVDS compatible DAC's

 

 

https://docs.google.com/spreadsheets/d/1h5PMUBkldkpt1rCnAR4ZHYGZNeCe-vwIFyKWYMZWsX0/htmlview

[ElviaCaprice]

The world is going to Ethernet renderers anyways. So won't be long before USB is phased out. This is a far better solution for those who want to use their general purpose computers as servers.

 

We can't even get folks from using YouTube over their phones, far from USB being phased out. In fact, I see USB audio being the meat and potatoes for the foreseeable future, if not already. Plenty of room for improvement on the audiophile end (see Regen). It's not just about SQ, it's also cost and user ability with what exists on the market today.

What I like about USB is it's universal acceptance and freedom of application use that is allowed today.

[Blizzard]

We can't even get folks from using YouTube over their phones, far from USB being phased out. In fact, I see USB audio being the meat and potatoes for the foreseeable future, if not already. Plenty of room for improvement on the audiophile end (see Regen). It's not just about SQ, it's also cost and user ability with what exists on the market today.

What I like about USB is it's universal acceptance and freedom of application use that is allowed today.

 

If far better Ethernet renderer solutions weren't coming out, what you say may be true. The next generation Ethernet renderers will put USB to shame. It's like putting the computer right into the DAC and connecting it directly to the DAC chip with very short PCB traces. Except the computers are purpose built for audio, powered by ultra low noise power supplies, and synchronously clocked by the the ultra low phase noise master clock the DAC uses. And they are the same footprint of today's USB interfaces.

 

And they won't be limited by UPNP/DLNA. If they are plugged into the same network your computer's on, they will pop up on the list of sound devices just the same as USB DAC's

 

And not to mention bandwidth. Up to DSD 1024 over 2 channels or DSD 256 over 8 is possible to a single endpoint.

 

 

Great example of a early gen version of these can be found in this DAC:

 

http://nadac.merging.com/networking

[firedog]

I think of I2s mostly as an interface between upmarket transport/DAC pairings from the same manufacturer. Excuse the dumb question, but is it a universal standard, so that if I have an I2s input DAC, it will work with any I2s output source, presuming they are compatible as to audioNo. format?

 

No, no universal standard for connector. There are a few different connectors in the market place. RG-45, HDMI, etc

[4est]

Have you spent any real time with one of these? It seems to me, USB was highly touted at first and here we are. For that matter, have you ever run the same DAC with i2s and another source concurrently in your own system? I2s is not new, nor is it obscure. The chips are made to use it, and it has become the standard over something like EIAJ. The problem is, and always has been, that it is attempting to send rf down wires without rf issues. Somehow, simply putting the sensitive rf signals in the same box as an rf source (the rendering cpu) does not appear to me to be a magic solution. As with everything, it is implementation, and we will see how it all shakes out.

If far better Ethernet renderer solutions weren't coming out, what you say may be true. The next generation Ethernet renderers will put USB to shame. It's like putting the computer right into the DAC and connecting it directly to the DAC chip with very short PCB traces. Except the computers are purpose built for audio, powered by ultra low noise power supplies, and synchronously clocked by the the ultra low phase noise clock master clock the DAC uses. And they are the same footprint of today's USB interfaces.

 

And they won't be limited by UPNP/DLNA. If they are plugged into the same network your computer's on, they will pop up on the list of sound devices just the same as USB DAC's

 

And not to mention bandwidth. Up to DSD 1024 over 2 channels or DSD 256 over 8 is possible to a single endpoint.

 

 

Great example of a early gen version of these can be found in this DAC:

 

NADAC | NETWORKING

[Blizzard]

Have you spent any real time with one of these? It seems to me, USB was highly touted at first and here we are. For that matter, have you ever run the same DAC with i2s and another source concurrently in your own system? I2s is not new, nor is it obscure. The chips are made to use it, and it has become the standard over something like EIAJ. The problem is, and always has been, that it is attempting to send rf down wires without rf issues. Somehow, simply putting the sensitive rf signals in the same box as an rf source (the rendering cpu) does not appear to me to be a magic solution. As with everything, it is implementation, and we will see how it all shakes out.

 

I'll be running this technology through the paces soon. Pretty much everyone is working on some sort of Ethernet renderer to put in their DAC's as we speak. It's a superior way to do things and the way things are going. These guys aren't fools. They aren't about to implement new technology that's inferior to the old technology.

 

Like I mentioned in my last post. The Merging NADAC is a good example of a early version of this technology.

 

The MSBtech network renderer is along similar lines, only problem is it's bound by UPNP limitations. But jitter wise, quite similar.

[vortecjr]

I'll be running this technology through the paces soon. Pretty much everyone is working on some sort of Ethernet renderer to put in their DAC's as we speak. It's a superior way to do things and the way things are going. These guys aren't fools. They aren't about to implement new technology that's inferior to the old technology.

 

Like I mentioned in my last post. The Merging NADAC is a good example of a early version of this technology.

 

The MSBtech network renderer is along similar lines, only problem is it's bound by UPNP limitations. But jitter wise, quite similar.

 

I'm not against what you are saying. However, here is what I know. I have contacted a lot of DAC manufacturers trying to promote the ethernet input idea and they generally seem uninterested. I offered them the Rendu's core ready to use so implementing things is pretty straight forward. One company jumped on the opportnity right away and two smaller companies are moving much slower to implement the Rendu's core.

 

Jesus R

[Jud]

Have you spent any real time with one of these? It seems to me, USB was highly touted at first and here we are. For that matter, have you ever run the same DAC with i2s and another source concurrently in your own system? I2s is not new, nor is it obscure. The chips are made to use it, and it has become the standard over something like EIAJ. The problem is, and always has been, that it is attempting to send rf down wires without rf issues. Somehow, simply putting the sensitive rf signals in the same box as an rf source (the rendering cpu) does not appear to me to be a magic solution. As with everything, it is implementation, and we will see how it all shakes out.

 

When you read about sending I2S between boxen with LVDS, you see mentions of RF issues, and eye tests, and jitter, and cable differences.... So wayyy different than USB.

[r_w]

IMHO.

 

My ears are telling me that USB is up to the task when implemented well, in a well optimised system.

 

My Aurender N100 and Auralic Vega (with regen and optical LAN isolation) outperforms my Linn LP12 and related fancy cart / phonostage (now redundant and removed from the lounge) in every department.

 

My guess is that if you can already pass the 1 & 0's without electrical and RF noise, in a nice reliable flow, the rest of the system is probably doing more to degrade the signal (Auralic seem to think that USB is their best input on the Vega).

 

I like the idea of using better clocks but I don't at the moment see why USB (done well) shouldn't be able to deliver spectacular music, as it already appears to be doing in my system.

 

Wouldn't Lumin be an example of renderer and DAC in one box? I don't hear of many people saying they are the ultimate way to hear music, just that they are good.

 

Sorry off topic for a sec, to context the statement above... being an ex record producer I remember an article in 'Sound on Sound' where almost every piece of pro audio kit using an external super clock had increased measured distortion over the internal one, but some did sound smoother to them. They seemed to think that nearest proximity to the DAC chip was an advantage, so it seems to be overall implementation that ultimately drives SQ, not just the transfer method.

 

Just my 2p... I'm not knowledgeable with the depths of digital audio, some on here obviously are, but I do think that perfect theory on paper rarely results in perfect outcome of those good intentions.

 

 

;-)

[Jud]

I do think that perfect theory on paper rarely results in perfect outcome of those good intentions.

 

 

;-)

 

Gawd, shouldn't that be posted above the figurative gates to this and every other audiophile site....

[Superdad]

Your wrong about the clock slaving. Most implementations don't even send the master clock from the source. The data gets clocked by the master clock when it gets into the DAC.

 

Excuse me? Do you not see the audio fs clocks in every I2S source? As well as the clock wire? The data goes to the DAC chips with those clocks and the DACs own clocks don't get used. Kind of the point with I2S--except that the external clock picks up all sorts of jitter on its way in.

(AFAIK, only MSB offers an I2S input with a clock line out to slave the source.)

 

Also notice how all the external I2S sources (Signature Rendu, OffRamp5, HydraZ) have expensive low- jitter clocks? It ain't for nothing...

 

If you are going to keep extolling the wonders of I2S, then please at least read up on how it works! https://www.sparkfun.com/datasheets/BreakoutBoards/I2SBUS.pdf

[4est]

Don't forget his DAC is Sabre and resamples. Resampling on input makes much of this moot anyway. It does use the local 100MHz DAC mclk.

Excuse me? Do you not see the audio fs clocks in every I2S source? As well as the clock wire? The data goes to the DAC chips with those clocks and the DACs own clocks don't get used. Kind of the point with I2S--except that the external clock picks up all sorts of jitter on its way in.

(AFAIK, only MSB offers an I2S input with a clock line out to slave the source.)

 

Also notice how all the external I2S sources (Signature Rendu, OffRamp5, HydraZ) have expensive low- jitter clocks? It ain't for nothing...

 

If you are going to keep extolling the wonders of I2S, then please at least read up on how it works! https://www.sparkfun.com/datasheets/BreakoutBoards/I2SBUS.pdf

[Blizzard]

When you read about sending I2S between boxen with LVDS, you see mentions of RF issues, and eye tests, and jitter, and cable differences.... So wayyy different than USB.

 

Where do you read about this? Never heard any of these reports about MSBtech, Sonore, or PS Audio's implementations. If you have please share the links.

[Blizzard]

IMHO.

 

My ears are telling me that USB is up to the task when implemented well, in a well optimised system.

 

My Aurender N100 and Auralic Vega (with regen and optical LAN isolation) outperforms my Linn LP12 and related fancy cart / phonostage (now redundant and removed from the lounge) in every department.

 

My guess is that if you can already pass the 1 & 0's without electrical and RF noise, in a nice reliable flow, the rest of the system is probably doing more to degrade the signal (Auralic seem to think that USB is their best input on the Vega).

 

I like the idea of using better clocks but I don't at the moment see why USB (done well) shouldn't be able to deliver spectacular music, as it already appears to be doing in my system.

 

Wouldn't Lumin be an example of renderer and DAC in one box? I don't hear of many people saying they are the ultimate way to hear music, just that they are good.

 

Sorry off topic for a sec, to context the statement above... being an ex record producer I remember an article in 'Sound on Sound' where almost every piece of pro audio kit using an external super clock had increased measured distortion over the internal one, but some did sound smoother to them. They seemed to think that nearest proximity to the DAC chip was an advantage, so it seems to be overall implementation that ultimately drives SQ, not just the transfer method.

 

Just my 2p... I'm not knowledgeable with the depths of digital audio, some on here obviously are, but I do think that perfect theory on paper rarely results in perfect outcome of those good intentions.

 

 

;-)

 

You can't just judge a complete product based on what interface it uses. There's many variables that make a product better than the other. Yes good USB implementations can sound good, but if you find a product that has both options, the I2S option will be the best sounding. This is no secret. Call up any manufacturer that makes a DAC/transport combo with both USB and I2S and ask them which the preferred method to connect them is. I can guarantee they will say I2S.

[Blizzard]

Excuse me? Do you not see the audio fs clocks in every I2S source? As well as the clock wire? The data goes to the DAC chips with those clocks and the DACs own clocks don't get used. Kind of the point with I2S--except that the external clock picks up all sorts of jitter on its way in.

(AFAIK, only MSB offers an I2S input with a clock line out to slave the source.)

 

Also notice how all the external I2S sources (Signature Rendu, OffRamp5, HydraZ) have expensive low- jitter clocks? It ain't for nothing...

 

If you are going to keep extolling the wonders of I2S, then please at least read up on how it works! https://www.sparkfun.com/datasheets/BreakoutBoards/I2SBUS.pdf

 

Maybe you need to read up on how I2S over LVDS works. Or maybe have a chat with John or Jesus.

[Blizzard]

Don't forget his DAC is Sabre and resamples. Resampling on input makes much of this moot anyway. It does use the local 100MHz DAC mclk.

 

As with most I2S over LVDS implementations, the master clock in the DAC clocks the data coming in. The Mirus clock is 50Mhz not 100.

 

The reason DAC's that have this option have it, is not to offer an inferior option to connect to their DAC's. It's to offer a superior option.