lionelag
lionelag's picture
Offline
Last seen: 4 years 1 month ago
Joined: Apr 18 2007 - 9:40am
USB or optical out? (Q for JA?)
struts
struts's picture
Offline
Last seen: 2 years 9 months ago
Joined: Feb 1 2007 - 12:02pm

Hi Lionel,

Guess JA is tied up at CES but you'll find the answer spread around the site. Bottom line is it depends how the USB DAC is implemented.

JA compared the jitter performance of Toslink and USB in his measurements of the dCS Scarlatti DAC in the August 09 issue. In this case while the USB source was a Macbook the Toslink source was a PC with a PCI soundcard, however I think it is still probably close enough to the scenario you describe to be a good indicator. The measurements show that in this case an audio-optimized USB interface (i.e. using asynchronous mode*) demonstrated considerably superior jitter performance to Toslink. Of course with a poorly implemented USB interface the situation could be reversed as it is with the Grace m902 as shown by JA's measurements thereof.

Afaik the only currently available USB DACs with audiophile pretensions that implement asynchronous mode are those from dCS, Wavelength Audio and Ayre. Implementing an audiophile-grade wordclock connection on top of asynchronous mode is non-trivial and requires computer expertise well outside the traditional audio domain. Up to now this has restricted the number of vendors able to offer high quality USB DACs, certainly at the budget end of the market. However Gordon Rankin has recently started licensing his 'Streamlength' code to other manufacturers so hopefully the choice of high quality USB DACs will start to increase quickly.

Please note that in both the examples cited here Toslink was not being compared with USB per se. One implementation of Toslink was being compared with one implementation of USB, in the same DAC in order to minimize the number of variables. As the differing results hopefully prove, it is not really possible to generalize up to the interface level based on these observations.

* There are of course other approaches such as reducing jitter downstream of the interface with SRC which, as Fig. 15 in the aforementioned Scarlatti measurements shows, can also be extremely effective. The Benchmark DAC1 USB is an example of a quite reasonably priced product that uses this approach although JA's measurements showed its USB jitter level to be higher than when connected via Toslink (although both were pretty low) indicating that the jitter level pre-SRC was also higher over the USB connection.

Log in or register to post comments
-->
  • X