some very interesting info here on the various ways the digital signal can be handeled 
, it gave me a clearer veiw of things that are very confusing to many of us.
The LessLoss info is good, but why is clock jitter the only consideration? Why is anyone concerned about jitter? Compared to other anomalies in digital audio, jitter is insignificant. Just look at the specs.
Jitter is the deviation in the precise timing of the clock signal that changes the output of the DAC chip with the value of the next sample. In simplest terms, sometimes the output of the DAC chips changes a little too soon or a little too late, and for those brief periods the DAC is outputting the wrong value. That timing error creates distortion in the reconstructed analog signal. For most listeners, 200ps of clock jitter is the threshold of audibility.
The CS841x series of digital interface receivers are quite common in DACs at all price points, from DIY to the bleeding edge. Jitter in the recovered clock is specified as 200ps rms. Compare that with the settling time of a typical multi-bit DAC chip.
Settling time is the time it takes from the moment the DAC chip gets the signal to output the new sample value, for the output current to reach the desired level, within +/- one LSB. During that time, the DAC is outputting the wrong value. The TDA154x series of DAC chips are also quite common in DACs at all price points. Settling times range from 200ns to 1000ns; more than 1000 times larger than jitter, yet no one seems to hear it.
This
oscillograph shows a TDA1543 in action. There is a lot going on because I wanted to show all the relevant signals in one picture. The green trace is the word clock, WS. The high to low transition of WS indicates the DAC is about to output the next sample value. For the TDA1543, that event happens on the next rising edge of the bit clock, the yellow trace. The blue and red traces are the left and right channel analog outputs. As you can see, the analog output levels begin to change coincident with the relevant clock, but it takes some time for the output to settle at its final level for that sample period.
The signal is a -8dB, 20KHz square wave with a 180-degree phase difference between the two channels. The shape of the settling curve looks like a classic, exponential time constant but it is much longer than the 500ns spec. The reason is that the spec was obtained by measuring sample chips under ideal conditions whereas the chip I am measuring is part of a real-life DAC component: a Lite DAC-AH, to be specific.
This
oscillograph is a 0dB, 20KHz sine wave with a 90-degree phase differential. It shows that settling time is not always a nice exponential. While it is settling, the output of a DAC is unpredictable and may include over/under-shoot and oscillation. It all depends on the size of the step, which current switches are changing, and the unique characteristics of the DAC chip. I believe the differences you hear between different DAC chips is largely due to the different output settling characteristics.
To make the comparison between jitter and settling time painfully obvious, near the left edge of the oscillographs are two (vertical) dashed lines; one is aligned with the rising edge of the bit clock that signals the DAC to output the next sample value, and the other is 200ps later. The space between the two lines, less than the width of one pixel on the display, represents typical jitter.
Steady-state sine waves are not the same as music. Because the signal representing music is forever changing, it is difficult to capture specific events with an oscilloscope. What’s more, we can hear aberrations many orders of magnitude smaller than we can see with a ‘scope. Although the test signals I used are not representative of real music, they dramatize what is happening with real music.
