Understanding Jitter in PC Audio

Mapman wrote:
"So in the case of jitter, once the clocking of the data is hosed, how practically does re-clocking it make it right again or at least better. What algorithm is used? Is it that the correct clocking is just implicit in the sample rate of the bitstream assuming all the original bits are transmitted? If so, why bother clocking data in these crazy digital audio systems in the first place?"

The data samples themselves have implicit in them the rate at which the recording samples were taken at record time. This is the standard, a perfect 44.1K samples per second for CD data. The reality is that the sample-rate at record time has jitter too, so it is not a perfect 44.1kHz. Recording studio equipment varies in quality and some A/D clocks are better than others. These sample inaccuracies at record time cannot be corrected. The samples are the samples, that's all you have.

The sample-rate inaccuracies at playback time on the other hand can be minimized. The sample-rate clock is part of the datastream only when the data is being transferred on playback, not when it is stored in RAM or on hard disk. It is the rate at which the sample data is transferred from one point to another, like from a Squeezebox to a DAC or from a Macbook Toslink to a DAC. Each tick of the clock transfers another bit from point A to point B. Without this clock, there would be no playback. The rate is set by the "source" device internal clock, such as the clock inside the Macbook or the clock inside the Squeezebox.

Reclockers goal is to throw away the incoming clock and totally replace it with a local oscillator at the same frequency. The problem is that frequency of the local clock will never be exactly the same as the frequency coming in unless the source clock is generated as a result of the reclocker local clock. This case is precisely what happens when you connect a word-clock cable from a reclocker to a Lynx Card for instance. The Lynx card adjusts its master clock to match the average frequency of the word-clock from the reclocker. Then, the average rate of the data from the source matches the average rate of data transfer of the reclocker output. If these match, then there is no data overrun, no lost bits, no drop-outs. I say "average data rate" because the actual data coming out of the reclocker does not match the incoming rate from clock to clock, only over a large number of clocks. This is what allows the reclocker to output a clock with lower jitter. The time interval between ticks of the reclocker output clock is more consistent than the time interval between ticks of the incoming stream clock. More consistent time interval equals lower jitter. Same thing.

Why reclock at all? Because jitter at the clocks of the D/A converter causes modulation of output analog signal from the D/A converter. This is distortion. This modulation is a function of both the magnitude of the jitter and the spectra of the jitter. This is one of the things that makes digital audio sound "digital" and not analog, along with sample rates that are not high enough.

The evidence of this is really obvious when you compare several DAC's to one another. With a high-jitter input signal, they all tend to sound radically different. With a low-jitter digital input signal, they all start sound very similar. Each DAC behaves a bit differently in the face of jitter, the simplest ones tending to sound the worst with high-jitter input and the best with low-jitter input.

Steve N.
Empirical Audio