reclocking

Steve,

I'm not comlaining about "cheap" USB converters, I complain about ALL USB converters. USB uses packets of data-music is best served via data streams. Pretty simple. No way to get around it.

Cerrot, what matters is the jitter at the point where the data stream is converted to analog. Whether or not the data is packetized when it is transmitted from the source component to the component in which that conversion is performed has no necessary relation to that. And, assuming comparably good implementation of the two approaches, packetized transmission stands a good chance of being better in that respect.

To achieve minimal jitter at the point of D/A conversion the clock used at that point should be independent of the timing of the signal that is transmitted between components. Otherwise performance will be degraded by jitter in the signal provided by the source; by noise, reflections, and waveform degradation occurring in the interface; and by clock extraction circuitry at the receiving end.

Asynchronous processing of packetized data inherently avoids those issues, or at least greatly minimizes them, since the timing of D/A conversion is essentially independent of the timing of the packetized data. Whether or not it may have other issues, depending on the specific implementation, is of course another matter.

Regards,
-- Al

Cerrot, I started this thread to point out that what ever is the technique used to get rid of the jitter at the end the data have to be switched to analog so the heart
of the system is the quality of the D/A converter not the interface.

Cerrot - Almarg is correct. Packetization has no bearing on the jitter in networked or USB systems. Both are buffered and master clock established in the device interface.

The older Adaptive technology relied on the computer clock, so it was deficient is this way. IT could only be improved with local PLL's which are no match for a free-running clock.

Likewise, DACs that have master clocks that are synchronized to an incoming stream are no match for a free-running clock. Jitter will be higher.

Buffering of the data and "pull" protocol combined with a free-running master clock with low jitter is simply the best way to achieve the lowest jitter in digital. See these plots to realize how low it can be:

http://www.audiocircle.com/index.php?topic=116570.0

The plots at the bottom approach the lower limit of jitter using current affordable technology. Most of the jitter is in the 15-18 psec range. You will not find a transport this low.

Steve N.
Empirical Audio

Alfe - you may think that the heart of the digital system is the DAC, but it isn't. It IS the interface that contains the master clock. That can be a built-in USB interface, external USB interface or a device and reclocker.

The problem is that most USB DACs have poor implementations of USB interfaces. The fact that the power and ground of the DAC USB interface shares these resources with other DAC circuitry is bad. This is what makes an external USB converter with a good power supply far superior to the majority of built-in USB interfaces. Just read the reviews. Most of them say that the S/PDIF input performs better than the USB interface.

Steve N.
Empirical Audio