Interesting. I gave up on transports many years ago. It doesn't take much to understand that physical spinning a disc live, then sending that data to a DAC has to have correction circuits and timing error issues, not to mention data read issues as well. If we send data from a file sat on an SSD there is no moving parts, so that in theory gets that problem out of the way.
As I see it, the sending of the data from a music server or PC / Mac is the next problem, how to get it with zero latency, and give the DAC the best chance of reading it.
And avoiding sending noise from the server to the DAC.
This is another ideal place to insert a Synchro-Mesh. Low jitter and galvanic ground isolation.
Then the DAC uses it's on board re-clocking / buffer to correct any additional jitter incurred in the sending process.
That would be nice in a perfect world, but that's not the world we live in. Input jitter still matters and the master clock inside the DAC doing the re-clocking is likely inferior to the one in the Synchro-mesh, so you are better off with a DAC that has no resampler on the input.
So far, I find sending via Ethernet sounds the best in my system, and my end point is SPDIF.
Likewise. That is why I stopped USB designs and designed an Ethernet interface. This is the best sounding interface I have offered yet.
Steve, what is your preferred entry protocol to the DAC? Is it SPDIf, AES, I2S? I am talking Redbook or 96K~ data rates, I don't use higher and don't upsample in my server, plus my DACs are NOS.
Thoughts?
Each type of input has its advantages and limitations:
S/PDIF - This can have very low jitter and galvanic isolation can be added at the driving or receiving end (it does take a small hit on jitter though). The downside is that some DACs don't use high-performance S/PDIF receivers, so the PLL in these can add audible jitter. I use the EK4114 because it adds insignificant levels of jitter. The 75 ohm cable and connectors can have a very good impedance match, particularly if it is a BNC terminated cable with 75 ohm RCA adapters.
AES/EBU - This is a higher voltage interface, so it must have an additional stage to get that higher voltage. This adds jitter. Minimizing stages is a key element in minimizing jitter. There is also the issue of the XLR connector not be matched impedance of 110 ohms. Because it is differential, it has the advantage of some common-mode noise rejection, but can also be galvanically isolated with a pulse transformer (adds jitter). It is my least preferred interface for these reasons.
I2S - This is a direct connection to the D/A logic, so it has the potential for lowest added jitter. The problems are that it can not be galvanically isolated without adding significant jitter and the cable in order to match the performance of S/PDIF must be quite expensive. If you are talking HDMI I2S like PSAudio, this is more strictly defined and terminated. If you are talking single-ended I2S on RJ-45 or the like, this can vary a lot as to termination technique, cabling and pinouts. I use both HDMI I2S and RJ-45 SE I2S. They sound great, but only marginally better than my S/PDIF, so I usually use S/PDIF myself. I think my S/PDIF cable is better than my silver I2S cable anyway.
Steve N.
Empirical Audio