USB: the lesser of the two evils?

[Coda II]

Current attention on USB for audio seems to be focused on the jitter thing and async applications to address this. But what about basic old fashioned electrical noise?

There are efforts to address this as well, but from a straightforward, non-technical pov, which is actually the lesser of these two evils?

 

[RobHolt]

Current attention on USB for audio seems to be focused on the jitter thing and async applications to address this. But what about basic old fashioned electrical noise?

There are efforts to address this as well, but from a straightforward, non-technical pov, which is actually the lesser of these two evils?

I think that the current crop of usb interfaces can be seen to improve matters technically - no question. I'm not convinced yet on the argument that this elevates usb above say optical given that the data is usually re-clocked in the dac. Plus there is still little in the way of listening tests to determine just how much jitter is acceptable before it becomes audible.

I like optical for the reasons you give and that's what I usually use.

 

[robert_cyrus]

potential spammer reported

 

[Telkman]

Cheers Robert, user has been banned.

 

[Tenson]

. Plus there is still little in the way of listening tests to determine just how much jitter is acceptable before it becomes audible.

There is this paper: Detection threshold for distortions due to jitter on digital audio

It places the audibility threshold around 250ns.

Listeners in the second experiment were all professionals, audio engineers,
recording/mixing engineers, musicians, etc... Sound materials were selected
by the listeners so that each listener could use his (her) familiar
materials. The experimenter (we) visited the listeners' studios or listening
rooms so that we could use listeners' own DAC, amplifiers, loudspeakers and
headphones. The system configurations, therefore, varied among listeners.

 

[Labarum]

Using my Beresford Caiman DAC I cannot here any difference between inputs USB, S/PDIF on Toslink or S/PDIF on phono plug.

 

[Tenson]

I'd have to say electrical noise is probably the more concerning factor to me. I've connected my external soundcard to a few computers and on occasion a system will put high frequency noise on the audio outputs. You can usually hear this change with what you are doing on the computer like accessing the hard disk. It's low, but it is audible if the soundcard is connected directly to a power amp.

On the other hand, random jitter must be at least 250ns to be audible which is much more than most systems produce these days. The worst case of sinusoidal jitter that I've seen, produced side bands about -115dB below the fundamental and only about 300Hz away from it, so masking effects would be huge. Other distortions like HD and IMD will be at least equal in amplitude and further spaced from the fundamental so closer to the threshold of audibility.

 

[Silk]

Current attention on USB for audio seems to be focused on the jitter thing and async applications to address this. But what about basic old fashioned electrical noise?

There are efforts to address this as well, but from a straightforward, non-technical pov, which is actually the lesser of these two evils?

My opinion, for what it's worth, is just because something could theoretically be audible, doesn't mean it is.

 

[sq225917]

I've had the Hiface, the Hiface Evo and the Young DAC in my set-up at various times.

In each case I've found them to be superior to the Optical out from my Macbook pro, I found the same was true of my previous dac as well. So I wouldn't assume all examples of a certain interface are as well implemented as each other.

I think each DAC responds differently to USB, some suffer with jitter, some suffer with noise on the power line and some are well designed, reclock or use an Async protocol and have carefully implemented grounding arrangement and well filtered power supplies.

The USB input of the Young is transformer decoupled, and doesn't use the 5v line at all so i guess that gives it a head start over some other USb dacs.

 

[felix]

On the other hand, random jitter must be at least 250ns to be audible which is much more than most systems produce these days. The worst case of sinusoidal jitter that I've seen, produced side bands about -115dB below the fundamental and only about 300Hz away from it, so masking effects would be huge.

[size=-2][ot:]
Julian Dunn put it at about 2-20nS; and the amount of jitter required for less than 1/2bit error to recover 16bit audio actually calculates at 246pS - a thousand times smaller. As you say, whether this is audible is very moot. Random stuff doesnt matter much, because it simply becomes noise. Correlated jitter is far worse, far more common, and more difficult to get a grip on. And there's some quite good evidence that it's the very tightness of that clock 'skirt ' which makes the real difference i.e. [low] frequency jitter <<100Hz - which does not agree with the published listening studies I've seen of random jitter injection[/ot][/size]

'Silk' above is spot-on. And its the whole package that matters, not the singling out of some nerdy figure of merit. No point having brilliant reclocking and a lousy/noisy output stage, or a tendency to hum, or other more gross defects, technical or aesthetic. And there's no arguing modern digital equipment that performs 'well enough' is a lot cheaper than it used to be.

I'd listen to a few dacs that do what I want within the budget and pick the one which pleased me most.

 

[Tenson]

[size=-2][ot:]
Julian Dunn put it at about 2-20nS; [/ot][/size]

Did he find that from listening tests, or educated speculation?

 

[sq225917]

Are you implying that you don't know his body of work?

Wow.

http://www.nanophon.com/audio/

A great man, sadly passed.

 

[Tenson]

Thanks for the link.

Some of it looks very interesting, but I don't have time to read all his work now! Can you please link to his listening tests on the audibility of jitter?

 

[felix]

Simon, I've got it in a book somewhere - will dig out & scan/send you an email this w/end.

All the Nanophon documentation and the technical work done for Audio Precision is well worth a read.

 

[robertz100]

I'm also using a Beresford Caiman, opted for TOSLINK because I wanted to avoid electrical interference (PC uses a high wattage PSU which the cable runs right past). It sounded a whole lot better than my USB trial using a printer cable but owing to the obvious quality difference in the two cables it is not really a fair comparison.

 

[jcbrum]

I don't understand what you mean here by the obvious quality difference ?

 

[sq225917]

I think he means that optical is galvanically decoupled form the power supply of the source and that USb is a great method for injecting switch mode noise into a dac via the 5v line.

Unless of course his USB cable was badly manufactured with bits hanging off it or something. ;-)

 

[RobHolt]

Bog standard USB is clearly noisy in most applications.
Look at the test results for both the MF Dacs and Cambridge and you'll see more noise and higher THD when using USB.

The only question is how audible these things are.
But since we have good alternatives (optical/coax/async usb) you might as well just not use standard USB where the above options are available.

 

[robertz100]

I don't understand what you mean here by the obvious quality difference ?

The usb cable was one originally chucked in with a printer, whereas the TOSLINK cable was a decent quality one costing £30+ a metre.

What I meant was I didn't compare the two methods like for like, I think a high quality audio usb cable would have given a fairer comparison. I'd assume the sound would be much better from a usb cable which used high quality silver-plated connectors and oxygen-free copper (probably costs 50x as much as the printer cable).

Having not heard the latter solution I can't comment beyond what I've read.

 

[sq225917]

The 'quality' of a USB cable will make no difference at all. If one cable had the 5v line isolated with a pulse transformer or similar than that would make a difference, but all the fancy construction in the world won't make any difference.

All the online usb cable bullshit stems from the Computer Audiophile site where they would rather waste £1000 on a cable than correctly isolate a £5 one.

There's 101 cable sellers on their in disguise pimping their wares with ZERO proof of any effect. Red what Steve Nugent from Empirical Audio says " use a correctly isolated USb input and the cheapest cable you can find".

Expensive cables are no substitute for good DAC design.