Data cabling
- Thread starter Powertech
- Start date
savatage1973
Addicted Member
Doesn't really matter--both are asychronous means of data transfer--USB is synchronous, which is better.
JoeESP9
ESL's & tubes since 83
IME it depends on the accuracy of the clock and associated circuits. With asynchronous the receiving device uses it's internal clock. This may be the better option especially with laptops. The best USB DAC's all operate asynchronously.
Gordon Rankin of Wavelength Audio wrote the first asynchronous protocols for USB. IIRC he got a patent for it.
Generally shouldn't matter with a good DAC, may matter on lesser DACs. Coax is considered to be better than Toslink. Quality gear should be using isolation transformers on coax to avoid noise/ground loop problems. Cables are cheap enough, try both and see if you can pick any differences.
I do NOT recommend to use Toslink or S/PDIF with any DAC that does NOT re-clock input stream. In both cases excessive jitter is unavoidable. With re-clocking or asynchronous rate converter jitter is completely eliminated.
House de Kris
Loud-n-Deep
Folks, both Coax and TOSLINK use the SPDIF protocol. So to say, "I prefer SPDIF to TOSLINK," doesn't really make sense. Saying you prefer Coax to TOSLINK makes sense. Saying you prefer copper to plastic makes sense. Saying you prefer electrical to optical makes sense.
I avoid Toslink. Yes, coaxial via BNC.
House de Kris
Loud-n-Deep
In my experience, jitter can never be completely eliminated. Heck, even passing a signal through a couple inches of PC trace has the potential of inducing an increase in measurable jitter. An SPDIF interface typically induces a "cyclical, correlated" jitter. Reclocking this stream will trade that flavor of jitter for "cyclical, uncorrelated" jitter. An asynchronous interface has the greatest chance of having the less benign form of jitter, "random," but it too will probably have a bit of cyclical uncorrelated jitter due to the power supply to the local clock.
O.K. I now have my Cambridge Audio 550C connected to a Behringer Ultramatch Pro upsampler. I have both coax and Toslink connected together with the original CA analogue outputs going to a spare preamp input foe comparison.
Direct comparisons-
1. I can't really hear any difference between the Toslink and the coax inputs to the Ultramatch, but there is noticeable difference when reclocking against synching directly to the input. This is the same with either input. Could it be that the Ultramatch effectively 'de-jitters' by default and that's why I can't hear any difference?
2. I can hear a difference between the direct analogue from the CD player and the Ultramatch for sure. The Ultramatch (using it's own DAC) sounds cleaner and smoother (I really hate audiophool speak, but this is the only way I can explain it).
I am now running the Ultramatch this way - I have now settled on using the coax link. Reclocking with it's internal clock and upscaling to 96KHz. This arrangement seems to give the best output. I will get (or borrow) an separate external DAC at some point and try this with and without the Ultramax in the mix.
The Ultramax also appears to work well with computer audio (Spotify) at 48KHz. I am using a Behringer UA202 (as a USB to optical converter) with it's optical link to the Ultramatch.
Anyway - so far, so good. I am quite impressed with the Ultramax - more than I thought I would be. I know that the comparisons I have made are in no sense any kind of blind testing, but I started out with a pessimistic view and I have tried to be as objective as possible. One further point is that the analogue output level is slightly higher than the CD player analogues and there is no digital level control inside it. I may investigate putting resistive attenuators in the line.
Direct comparisons-
1. I can't really hear any difference between the Toslink and the coax inputs to the Ultramatch, but there is noticeable difference when reclocking against synching directly to the input. This is the same with either input. Could it be that the Ultramatch effectively 'de-jitters' by default and that's why I can't hear any difference?
2. I can hear a difference between the direct analogue from the CD player and the Ultramatch for sure. The Ultramatch (using it's own DAC) sounds cleaner and smoother (I really hate audiophool speak, but this is the only way I can explain it).
I am now running the Ultramatch this way - I have now settled on using the coax link. Reclocking with it's internal clock and upscaling to 96KHz. This arrangement seems to give the best output. I will get (or borrow) an separate external DAC at some point and try this with and without the Ultramax in the mix.
The Ultramax also appears to work well with computer audio (Spotify) at 48KHz. I am using a Behringer UA202 (as a USB to optical converter) with it's optical link to the Ultramatch.
Anyway - so far, so good. I am quite impressed with the Ultramax - more than I thought I would be. I know that the comparisons I have made are in no sense any kind of blind testing, but I started out with a pessimistic view and I have tried to be as objective as possible. One further point is that the analogue output level is slightly higher than the CD player analogues and there is no digital level control inside it. I may investigate putting resistive attenuators in the line.
JoeESP9
ESL's & tubes since 83
If you have an asynchronous USB DAC the jitter level is dependent on the quality of the clock in the DAC.
asynchronous USB DAC Definition from PC Magazine Encyclopedia
Definition of: asynchronous USB DAC. asynchronous USB DAC. A digital-to-analog converter (DAC) used for music playback that connects to the computer via USB. The asynchronous USB DAC (async USB DAC) uses its own clock to regulate the data rate from the computer, which allows for jitter-free data transfer.
asynchronous USB DAC Definition from PC Magazine Encyclopedia
Definition of: asynchronous USB DAC. asynchronous USB DAC. A digital-to-analog converter (DAC) used for music playback that connects to the computer via USB. The asynchronous USB DAC (async USB DAC) uses its own clock to regulate the data rate from the computer, which allows for jitter-free data transfer.
cpt_paranoia
Addicted Member
I blame the person who coined the 'asynchronous' name for a destination-clocked transfer scheme as opposed to a source-clocked transfer scheme.
SPDIF is a unidirectional interface, with a simple data protocol; data flows only from source to DAC. Therefore, the data source is the clock master for DAC operations. There is no way for the destination to perform flow control of the source.
USB is a bidirectional interface, supporting a wide range of data protocols. It can support a source-clocked protocol, or a destination-clocked protocol.
A destination-clocked protocol is better, since the DAC clock can be the master clock. This can be very clean. It requires a protocol to cope with the different clock mechanisms between the DAC and the data source, which requires FIFOs and associated content detection, and packet requests or flow control from destination to source when more data is required. There is no short-term relationship between the source data clock and the sink (DAC) data clock.
A source-clocked protocol relies on the source clock to clock the DAC. Since this is often a PC, the clock is often noisy, and sometimes irregular. Without clock regeneration within the DAC, this protocol type can lead to higher jitter. Since a regenerated clock will never be as clean as a master clock, this transfer mechanism will never be as good as a destination-clocked protocol.
Within the DAC interface world, the term 'synchronous' has been applied to source-clocked protocols (possibly because the DAC clocking is synchronous with the source), and 'asynchronous' applied to destination-clocked protocols (possibly because the source and destination clocks do not need to be the same).
So, SPDIF is a 'synchronous' protocol, and USB can use either 'synchronous' or 'asynchronous' protocols.
(If we're going to be picky, the destination-clocked protocol is actually isochronous; nominally the same frequency, but supporting short-term variation between end points.)
Not quite. There is no such thing as a zero jitter clock. It's reduced to whatever the jitter performance of the retiming clock is, which can be sub ps these days, and can reasonably be as low as c.100fs RMS, which is an astonishing improvement over say, a decade ago.
I don't believe that is entirely correct. There is no clock line sent with the USB signal.
https://www.silabs.com/documents/public/white-papers/usb-audio-simplified.pdf
The above white paper from Silicon Labs explains the fairly complex requirements for USB isochronous transfers as used in Digital Audio.
