advertisement


USB audio coming of age?

Status
Not open for further replies.
I

ItemAudio

Trade: Item Audio
Finally, we're seeing a healthy number of USB-specific audio cards on the market, and a fair few enterprising folks clock- and power-upgrading their USB output for connection to DACs.

The latest SoTM, Aurender and Auralic are pitched much more squarely as USB specialists. And even Gigabyte has a range of motherboards with low-noise USB outputs specifically designed for this task.

Which have you found the most effective?

For some time I've been touting the idea that DAC manufacturers should release a matching PCIe card for dedicated USB output instead of just a driver, but you know how it is: no-one listens to me until the plagiarism kicks in five years later . . . but in the last ten years, USB has gone from a fringe connection method - something to be tacked on to a DAC to grant 'token' computer connectivity - to the pre-eminent protocol, carrying everything up to DSD128. Now it seems that attention is being focused (belatedly) on the forgotten partner in the source: the USB 'transport'
 
Why so few streamers with USB output? I couldn't find a single cheap one. In fact, AURALiC is all I found. I'm guessing your answer will be Mac Mini..
 
Most proprietary streamers (usually computers in disguise) are Linux-based - which is finally a viable proposition thanks to the increase in Class 2.0 implementations.

The £399 SoTM USB 'streamer', for instance, is cheaper than a Mac Mini.

But all the USB cards are PCIe based - so they play into Intel/Windows/Linux territory.
 
Alex S said:
Why so few streamers with USB output? I couldn't find a single cheap one. In fact, AURALiC is all I found. I'm guessing your answer will be Mac Mini..
Click to expand...
Because USB is not as simple & stupid as SPDIF. It's *much* more than simply streaming bits that make up discrete snapshots of amplitude values in deterministic time frames. It's a widely defined computer protocol that doesn't restrict the devices in ways SPDIF does - yes, there are some standards on how to do things (UAC), but these are only the baseline for the entire device-side implementation. There's so much that can go wrong, from device enumeration, to URB scheduling, to additional features (HID, volume control, ..) that these things do indeed break with non-windows-xp systems quite often (since winxp (and windows in general) needs workarounds on its own).

In the end, I can imagine it's not worth it for many manufacturers - if DAC XYZ, which works perfectly under windows, but doesn't work with the streamer out-of-the-box, the streamer gets blamed. Once again - it's not rare for the "not work" to occur - just take a look at how many workarounds ie. the Linux kernel has, just for handling the control interfaces of these external devices - there's a whole system dedicated for "quirks" or vendor-specific hacks just to get/set supported bit depth or sample rates correctly.

Just a few examples of the very few devices that have proper workarounds in Linux,
http://lxr.free-electrons.com/source/sound/usb/quirks-table.h
Now imagine a streamer vendor having to do something similar for audiophile DACs.

And I'm not even getting into the non-UAC DAC territory.

There - USB is convenient for the end user, but not really for the streamer manufacturer (assuming the streamer has a custom OS). :)
 
When it comes to audio, the U in USB is a cruel joke. That's why I'm hoping DAC manufacturers treat the upstream a bit more seriously.

Digital audio sources are considerably improved by companies like Auralic and dCS taking control of both ends of the transfer. Auralic went a step further than recommending JPlay - they built a USB transport for the Vega. Smart move: they're setting the pace.

However, SoTM, JCat, PPA and others currently under development - including a fair few DIYers - are also showing how a stock PC can be significantly improved with USB clock upgrades. I was reading today about a Phasure DAC's owner who 'synced' both ends of the protocol with Dexa clocks - replacing the one in the DAC and the one on the USB card. That's the way to do it right.
 
Digital audio data transfer is by its very nature, perfect.

How can this be improved, are you talking in terms of software, functionality, convenience, Mark?
 
I'm afraid you've lost me.

IMO, there's no such thing as 'USB Audio', there's a bunch of different ways that one can use digital audio and USB may be used, or not, and none of them offer better or worse sound quality using the same digital recordings/files.

It's really quite simple, but you seem keen to make digital audio seem anything but....
 
Max, ask yourself what the op does for a living, which explains why this thread, and why posted in the wrong area as usual.
 
Business needs a lift, few areas to exploit. Cables make deafening differences with digital signals. Turntable fettling minds will latch on to tuning/tweaking their transport so a ready prepared market.

I want an improvement so I look for and find one.

I am still trying to hear the amazing improvement a Wire World USB cable can make compared to any other USB cable.
 
It's simple, buy a decent dac with good noise rejection and decent asynchronous operation and be done with it. There's no currently valid testing showing the audibility of jitter and noise with USB dacs or any audible differences between different software players.
 
I was hoping you few chaps had been tutored by experience since the last USB chat! But it's still a USB present and future, for as long as two-box digital sources are the norm.

The newsflash from outside the cave is that DSD-capability is driving maturation of the interface DAC-side, and weight is gathering behind a more concerted push to address upstream issues.

If you're turning a deaf ear to all this development, the only contribution you can usefully make here is registering your intent to ignore it. Message now received - your vote is important, too. If you can, enjoy listening to your system.
 
maxflinn said:
I'm afraid you've lost me.

IMO, there's no such thing as 'USB Audio', there's a bunch of different ways that one can use digital audio and USB may be used, or not, and none of them offer better or worse sound quality using the same digital recordings/files.

It's really quite simple, but you seem keen to make digital audio seem anything but....
Click to expand...

Max,
The guy makes his living flogging solutions to non-existant problems. The more variables that can be squeezed into the chain, the more opportunities to spread FUD.

Digital audio is very, very simple. Essentially, it's about transferring data & converting that data into an analog waveform. And we've known how to do that with 100% reliability for several decades.

Chris
 
Obvious this is happening. The noise rejection(is there a test for this?) is occurring possibly by the independent powering of USB cards.
Diyers Item are struggling with clock upgrades. Powering the clocks independently/desoldering removing and soldering up the new clocks seems fraught with difficulty. PPang does it for a price but no guarantee given.
 
jirij said:
Because USB is not as simple & stupid as SPDIF. It's *much* more than simply streaming bits that make up discrete snapshots of amplitude values in deterministic time frames. It's a widely defined computer protocol that doesn't restrict the devices in ways SPDIF does - yes, there are some standards on how to do things (UAC), but these are only the baseline for the entire device-side implementation. There's so much that can go wrong, from device enumeration, to URB scheduling, to additional features (HID, volume control, ..) that these things do indeed break with non-windows-xp systems quite often (since winxp (and windows in general) needs workarounds on its own).

In the end, I can imagine it's not worth it for many manufacturers - if DAC XYZ, which works perfectly under windows, but doesn't work with the streamer out-of-the-box, the streamer gets blamed. Once again - it's not rare for the "not work" to occur - just take a look at how many workarounds ie. the Linux kernel has, just for handling the control interfaces of these external devices - there's a whole system dedicated for "quirks" or vendor-specific hacks just to get/set supported bit depth or sample rates correctly.

Just a few examples of the very few devices that have proper workarounds in Linux,
http://lxr.free-electrons.com/source/sound/usb/quirks-table.h
Now imagine a streamer vendor having to do something similar for audiophile DACs.

And I'm not even getting into the non-UAC DAC territory.

There - USB is convenient for the end user, but not really for the streamer manufacturer (assuming the streamer has a custom OS). :)
Click to expand...

Excellent summary: development of USB audio is caught between the juggernaut of the IT industry - with its terrifyingly low margins - and the puttering little caravan of boutiques that comprise the audio biz.

Ultimately, I suspect these stress fractures will be complicit in the demise of USB as an interface for audio, as tablet-controlled smart functionality finds its way into DACs, amplifiers and speakers . . . and within them, I2S.

But, for now, the drive to perfect both ends of the USB transfer is a necessary chapter in the story. Not long ago, no-one cared which USB chipset and clocks were used in the DAC; now they're headline features. That's progress.
 
tonerei said:
Obvious this is happening. The noise rejection(is there a test for this?) is occurring possibly by the independent powering of USB cards.
Diyers Item are struggling with clock upgrades. Powering the clocks independently/desoldering removing and soldering up the new clocks seems fraught with difficulty. PPang does it for a price but no guarantee given.
Click to expand...

Not obvious to some, apparently! But yes - there have been some interesting threads elsewhere in the last month about this.

Paul's 25MHz motherboard clock upgrades are fraught with problems - and for best results he's battery-powering them, which introduces another layer of grief. It's something we're looking at, too.

But it's much easier to linear-power the 24MHz or 48MHz unit on a £7 USB3 card: just desolder the through-hole clock and replace it with the equivalent XPress0 Ultra, Tentlabs, or - if, you're feeling wild, Dexa clock/board and you're away.

Doing this will be a) much more beneficial than modding the mainboard, b) easier and c) less risky - fewer tears fritzing a beer-money PCIe card!
 
ItemAudio said:
Not obvious to some, apparently! But yes - there have been some interesting threads elsewhere in the last month about this.

Paul's 25MHz motherboard clock upgrades are fraught with problems - and for best results he's battery-powering them, which introduces another layer of grief. It's something we're looking at, too.

But it's much easier to linear-power the 24MHz or 48MHz unit on a £7 USB3 card: just desolder the through-hole clock and replace it with the equivalent XPress0 Ultra, Tentlabs, or - if, you're feeling wild, Dexa clock/board and you're away.

Doing this will be a) much more beneficial than modding the mainboard, b) easier and c) less risky - fewer tears fritzing a beer-money PCIe card!
Click to expand...

This is a good example of why streamers are going to be around for a while longer.
 
Love streamers. Hate computers.

See also:*
Love animals; love meat.
Hate tomatoes; love tomato sauce.
Love Scotland; hate Scots.
Pro abortion; anti capital punishment.

------
* The views expressed in this post are not necessarily those of the author and do not necessarily represent the views of, and should not be attributed to, Item Audio or Pinkfish Media.
 
Status
Not open for further replies.


advertisement


Back
Top