A number of discussions that have referenced jitter. Some background for anyone who cares.
Picture yourself clapping along with a piece of music. Generally you clap in time with the beat, sometimes a little early, sometimes a little late.
If you think of the music beat as the ideal clock, and your clapping as the actual clock, then the differences between the two is the jitter. On average, your clapping is in time, but it may deviate from the correct beat in a random way. The technical term for this is phase jitter - the actual clock is the right frequency, but doesn't always tick at exactly the right time.
DACs can suffer from jitter in two ways:
1. In a synchronous system, the clock tick is the point at which data is considered valid. So for a S/PDIF data stream, every time the clock ticks the DAC looks at the S/PDIF data line and notes whether it is 1 or 0. If the clock has jitter, it is possible that the DAC looks at the S/PDIF data too early or too late and reads the data in error. These days, S/PDIF data rates are considered slow, and there is no system I know of that has so much clock jitter that data is read in error.
2. In a digital audio system, the clock tick is the point at which the audio sampled by the A2D converter. Implict in the process is that the clock ticks in the same regular way when the samples are played back by the DAC. If the clock used for the playback ticks slightly early or late (it has jitter) then the samples are played back slightly early or late and lead to distortion in the output signal. This is how jitter can effect the audio quality of the DAC output.
As jitter is a random process, its frequency / probability distribution impacts how audible it is. There is no standard way of reporting jitter, and it gets reported in different ways. I have seen peak jitter, average jitter, RMS jitter all reported - these figures are not comparable.
There have been a number of studies about the point jitter becomes audible, but no real consensus in academic papers. There have also been suggestions that jitter correlated to the audio signal is much more detectable than jitter that is purely random.
Picture yourself clapping along with a piece of music. Generally you clap in time with the beat, sometimes a little early, sometimes a little late.
If you think of the music beat as the ideal clock, and your clapping as the actual clock, then the differences between the two is the jitter. On average, your clapping is in time, but it may deviate from the correct beat in a random way. The technical term for this is phase jitter - the actual clock is the right frequency, but doesn't always tick at exactly the right time.
DACs can suffer from jitter in two ways:
1. In a synchronous system, the clock tick is the point at which data is considered valid. So for a S/PDIF data stream, every time the clock ticks the DAC looks at the S/PDIF data line and notes whether it is 1 or 0. If the clock has jitter, it is possible that the DAC looks at the S/PDIF data too early or too late and reads the data in error. These days, S/PDIF data rates are considered slow, and there is no system I know of that has so much clock jitter that data is read in error.
2. In a digital audio system, the clock tick is the point at which the audio sampled by the A2D converter. Implict in the process is that the clock ticks in the same regular way when the samples are played back by the DAC. If the clock used for the playback ticks slightly early or late (it has jitter) then the samples are played back slightly early or late and lead to distortion in the output signal. This is how jitter can effect the audio quality of the DAC output.
As jitter is a random process, its frequency / probability distribution impacts how audible it is. There is no standard way of reporting jitter, and it gets reported in different ways. I have seen peak jitter, average jitter, RMS jitter all reported - these figures are not comparable.
There have been a number of studies about the point jitter becomes audible, but no real consensus in academic papers. There have also been suggestions that jitter correlated to the audio signal is much more detectable than jitter that is purely random.