“Is I2S better than USB?”

“Will I benefit from adding a DDC (digital to digital converter) and connecting it to S5 or AYA 5 A-link input, instead of using their USB input? Is A-link better than USB?”

I am regularly receiving such questions. And here is the reply.
Somewhat more than a decade ago, I wrote several articles on different interfaces, and their advantages and shortcomings. The most comprehensive were two articles I wrote in 2008 and 2010, which dealt with performance and the main differences between USB and S/PDIF interfaces. In 2010, we came across the point where the clocking scheme for USB DAC can be no-compromise, and its performance consequently comes down to the solutions applied internally. So, the USB DACs are not supposed to have intrinsic limitations anymore. They are as good as their design is good.

Practically speaking, Audial DACs made since 2010 use quality clocks and clocking schemes for their USB part. The latest S5 DAC uses Crystek clocks, with phase noise of -100 dBc/Hz at 10 Hz, and the way they are used here they achieve even 5 dB better performance. While I am perfectly aware of some clock designs achieving even somewhat better figures, such a performance can be certainly considered extremely good. Mind you, it is way better than anything available up until 6 or 7 years ago.

As opposed to the USB interface, the A-link, being an electrical “shell” for I2S and similar internal protocols, provides practically a direct path to the TDA1541A chip. It means that the performance with this input is as good as the source is.

So, there is no answer to the question “what’s better”. If you believe that your (A-link compatible) I2S or possibly Philips simultaneous data protocol source sports better performance than the internal USB stage of the Audial DAC, the chances are it will be better.

Please however note that I can not answer the questions on the qualities of certain I2S sources, compatible with A-link, it is just not my part to test different units or comment on the other manufacturers’ products, sorry. If you believe in the advantages of some DDC, it is on you to proceed and check it.

To make this comparison complete, I should also mention that the latest S5 (and AYA 5) has also S/PDIF electrical and optical (“Toslink”) input. Both these inputs carry the same signal, the difference is in their form, which is either electrical (ones and zeroes are presented by the voltage going up and down) or optical (one and zeroes are the light turning on and off). Generally, the S/PDIF inputs are “slaves” which means that the clock is locked to the source, and hence these inputs’ performance also notably depends on the source. These inputs however can to some degree “attenuate” the incoming jitter, and the S5 offers two modes of operation in this regard, one recommended for use with good (“data lock”), and the other recommended for use with sub-optimal sources (“pre-amble lock”).

Normally, electrical S/PDIF is superior to optical, simply because it avoids conversion from the voltage to the light at the source, and then back from the light to the voltage at the DAC. And these optical transmitters and receivers are usually not the best jitter performers. Still, the S/PDIF can work well, even the optical interface can sound decent – I am regularly listening to the TV by using the optical connection between the STB and my DAC.

For now, on this website you could see the graphs showing the frequency domain performance of the USB and S/PDIF clocks used in the S4 DAC, and also the J-signal analysis of the S5 audio output, showing the difference in jitter performance of two S/PDIF modes of operation. To make these things more obvious, and show more direct comparisons, I plan to add some new graphs soon.



See also:

High resolution audio, asynchronous USB, oversampling, upsampling and stuff: The current state of affairs (2010)

Asynchronous USB (2010)

A-link (PCM / I2S direct) specifications

Master clock frequencies and jitter performance

Leave a Reply