AES vs USB

[zenpmd]

I understand that, in latency terms, AES/EBU will be at least twice as fast as USB. Is that correct? What other benefits are there to AES?

 

There is something about USB which just doesnt seem "right" to me.

[One and a half]

Compared to USB, packet noise from an AES3 transmission is a lot less, therefore impacting less on the power supply rails.

 

Cable length is 100m versus 2m with USB.

 

The connector is far more durable than USB will ever be although the size is about twice current laptop thickness.

 

AES3 does not carry VBUS or a signal ground. Only a differential pair and a case earth wire.

 

AES3 cards like the RME HDSPe AIO are PCIe only, therefore requires a motherboard with real slots in it and can't be used with a laptop, unless they have an e-sata connection and some god awful workaround like Thunderbolt.

 

IMHO AES3 trounces USB for the DACs I own, takes off the grunge and the music is engaging.

[Music Matters]

My experience is also more positive with AES3 than USB, I use Hiface EVO converter to NorthStar Supremo DAC.

[zenpmd]

Im thinking of getting a Lynx card which allows this. Does anyone have any experience of that?

[Audio_ELF]

I always think of it this way ... theoretically async USB interface direct (via i2s) to a DAC is going to be the "best" option. The problem is the real world doesn't work in theoreticals and so the additional isolation offered by going via AES (or SPDIF) overweighs the disadvantages of clocking issues.

[Audio_ELF]

Im thinking of getting a Lynx card which allows this. Does anyone have any experience of that?

For a long time Chris promoted the Lynx card as a good way to output AES to DACs such as the Berkeley, but I belief he feels the Berkeley Alpha USB is a better method these days.

[mordante]

USB is chosen because of convenience. Not out of suitability.

 

There have been lots of software problems using USB. As far as I know Microsoft still does not support USB audio 2.0.

[zenpmd]

USB is also a pathetically clumsy connection when you start getting used to XLRs as I now am!

[Fyper]

Im thinking of getting a Lynx card which allows this. Does anyone have any experience of that?

 

I'm still using a Lynx AES16 PCIe. Their drivers are still evolving and now you can have a player such as Foobar 2K using their latest 64 bit driver.

I'm quite happy about it. I use it as dual AES and synchronize the card and the DAC with a clock.

The whole set up has been in place for some years now, and I'm not sure there is something providing better results out there.

 

That being said, if you want to play 32/384 files or DSD no converted to PCM 24/192, you'll have to use USB.

And generally speaking, I have no doubt that you can have excellent results with nowadays USB.

[dean70]

AES is limited to PCM 24/192. I am surprised none of the manufactures have looked at extending the specification to support higher rates & native DSD.

[Allan F]

As far as I know Microsoft still does not support USB audio 2.0.

 

Not a concern. Although Microsoft may not, virtually every DAC manufacturer with USB input provides USB 2.0 drivers.

[Fyper]

AES is limited to PCM 24/192. I am surprised none of the manufactures have looked at extending the specification to support higher rates & native DSD.

Especially since we know, thanks to the existence of dual AES, that the band width is of at least 2X24/192 in dual wire and probably more.

The thing is, since USB does that already , why bother developing a new standard.

But even before that, the proprietary solutions developed by Esoteric, Denon, etc. to transmit DSD were never developed from an AES base.

[mordante]

AES/ebu is not the digital format which is under appreciated. AT&T optical (ST fiber) as used by Wadia, Bel Canto and some other brands is maybe one of the best digital connections. Too bad so few brands support it.

[barrows]

Remember, "AES" is just balanced SPDIF. Unless you are talking about video synching, latency has no relevance to audio playback.

SPDIF/AES both suffer from the same problem, the four lines of the I2S signal are combined into one signal, and receiving and de-embedding the I2S signal from the SPDIF/AES feed increases jitter.

Now, the problem with USB is that many DACs do not do not implement their USB receivers well enough at all, but, when, and if, USB is properly implemented, it will trounce SPDIF/AES in terms of jitter, and this is proven by proper jitter measurements as well.

 

SPDIF/AES has always been a compromised format, and I would love to see it disappear entirely.

[Miska]

That being said, if you want to play 32/384 files or DSD no converted to PCM 24/192, you'll have to use USB.

And generally speaking, I have no doubt that you can have excellent results with nowadays USB.

 

DSD64 works fine over single-wire AES (as does with S/PDIF too). For DSD128 you need dual-wire, but there won't be many DACs that support DSD128 over dual-wire AES...

[r_w]

USB is sounding fabulous to me, can't knock it when it's been optimised.

 

It doesn't matter to me that it needs optimising if the result of doing so is as good as I'm hearing, there is no point slating something that can sound so good.

 

These optimisation add-ons have highlighted areas where manufacturers can improve their implementations so it's probably going to get even better by default.

[Fyper]

[..]Now, the problem with USB is that many DACs do not do not implement their USB receivers well enough at all, but, when, and if, USB is properly implemented, it will trounce SPDIF/AES in terms of jitter, and this is proven by proper jitter measurements as well.

One could also say that, if AES is properly implemented (using efficient reclocking for example), jitter won't be worse than well implemented USB. And you certainly can measure that as well...

[Fyper]

DSD64 works fine over single-wire AES (as does with S/PDIF too). For DSD128 you need dual-wire, but there won't be many DACs that support DSD128 over dual-wire AES...

That was kind of my point in my original message: it's not a question of capacity but rather a question of avoiding SACD Ripping by allowing DSD over AES-S/PDIF.

I'd be curious (and interested) to know which DACs accept DSD over AES (64 or 128), or are you talking about DoP ?

[mordante]

Why talk about DSD?

 

DSD is such a small part of the audio market that it's really not that interesting to focus on.

 

Last weekend I went to a concert and talked to one of the musicians. He never even heard of DSD nor did his studio know it. He say all recording is done at 24/48 or sometimes 24/192 but all mixing and mastering is done at 16/44.1 since the end product will also be at 16/44.1.

 

He also said his best sounding album according to him was recorded on tape and all the mixing and mastering was also done analog. But it is too expensive for him, so he stopped working that way.

 

I have spoke with a few other musicians and people who work in studios. Most are not even remotely interested in hi res PCM and never even heard of DSD.

[Solstice380]

AES/ebu is not the digital format which is under appreciated. AT&T optical (ST fiber) as used by Wadia, Bel Canto and some other brands is maybe one of the best digital connections. Too bad so few brands support it.

 

+1 In 1995 I had a Theta Digital transport and DAC that had the ATT fiber and it was by far the best hookup on that combo. And the cable is relatively inexpensive compared to good digital coax cables. It's not even worth talking about Toslink!

[4est]

Although this can be true, it seems out of place to use an example that is re clocked. By definition, a reclock would make all inputs the same for better OR for worse.

One could also say that, if AES is properly implemented (using efficient reclocking for example), jitter won't be worse than well implemented USB. And you certainly can measure that as well...

[CatManDo]

He say all recording is done at 24/48 or sometimes 24/192 but all mixing and mastering is done at 16/44.1 since the end product will also be at 16/44.1

 

I think everybody is now using 24bit for mixing/mastering (sampling rate can be between 44.1 and 192 kHz). The downconversion to 16/44 is the final step after the mastering.

[accwai]

+1 In 1995 I had a Theta Digital transport and DAC that had the ATT fiber and it was by far the best hookup on that combo. And the cable is relatively inexpensive compared to good digital coax cables. It's not even worth talking about Toslink!

 

My main transport is a Krell DT-10. When paired with its matching Ref 64 DAC, it sends data and clock in separate AT&T links. According to a Stereophile review from that era, the AT&T links sound significantly better than AES between the two. I can't think of one current top tier DAC that supports AT&T fiber though. So AES will have to do for new stuff.

 

Although this can be true, it seems out of place to use an example that is re clocked. By definition, a reclock would make all inputs the same for better OR for worse.

 

Unfortunately, it isn't quite the case in practice. My setup is double reclocked--the DAC has a built in reclocker and there is an MC-3+ in front. But I can hear significant difference among transports and different types of links from the same transport. Toslink is always worst by a significant margin.

[firedog]

Why talk about DSD?

 

DSD is such a small part of the audio market that it's really not that interesting to focus on.

 

Last weekend I went to a concert and talked to one of the musicians. He never even heard of DSD nor did his studio know it. He say all recording is done at 24/48 or sometimes 24/192 but all mixing and mastering is done at 16/44.1 since the end product will also be at 16/44.1.

 

He also said his best sounding album according to him was recorded on tape and all the mixing and mastering was also done analog. But it is too expensive for him, so he stopped working that way.

 

I have spoke with a few other musicians and people who work in studios. Most are not even remotely interested in hi res PCM and never even heard of DSD.

 

Most musicians know little about recording/hi-end audiophilia and the various stuff people in this forum are interested in. In fact, it seems that in most cases the musicians have little control over how their music is produced, mixed, and mastered. So I don't know that what your musician friend thinks is very relevant here.

 

It's true that DSD is not widespread; but it has a fairly serious amount of use in recording classical music, so for classical lovers it is very relevant. Additionally, a lot of the members here like to upsample music to DSD for playback, as they find that's the best sounding result in their system. So for them DSD is relevant. This is a site for "audiophiles", afterall, not average Joe listeners or even "most people".

 

AFAIK, most music these days is recorded in some form of hi-res (24/44.1 or above) and generally there is hi-res mixing and a hi-res master made. Conversion to Redbook comes later. I think your friend was probably misinformed about how mixing and mastering is commonly done today. That's why we often see hi-res releases of even new music nowadays.

[Miska]

I'd be curious (and interested) to know which DACs accept DSD over AES (64 or 128), or are you talking about DoP ?

 

Yes, DoP is the method to carry it over. At least Mytek supports it, IIRC also Chord and Benchmark. Maybe some other too...