Measuring Jitter in AP2700

Created on 2008-07-11 13:13:00

Jitter is the variation in time of the derived clock signal from nominal. Jitter can be introduced into a digital audio signal during the sampling process, and/or by the digital interface. The three types of jitter are "sampling jitter" (induced by an ADC's imperfect clock), "sync jitter" (induced by a receiver's VCO imperfection), and "interface jitter", where cable reactance and/or improper impedance cause a loss of high frequency, thus smearing the clock's pulse transitions. A digital audio system relies on sample points at specific locations in time. Jitter results in distortion of the reproduced audio stream within the digital stream as sample points are shifted forward and backward in time.

The 2722 Series audio analyzer offers three methods of jitter detection using the Quick Launch macro Intervu. Intervu analyzes the AES/EBU or consumer serial digital interface input signal via an 8-bit A/D converter with an 80 MHz sample rate, providing an analysis capability with over 30 MHz bandwidth. Intervu acquires 1.5 M samples of the interface signal into a buffer, resulting in about 19.66 milliseconds of data.

The Preamble jitter detection selection uses the average rate of the trailing edge of the first three-UI-wide pulse in each preamble as the stable clock reference. Each actual transition at the trailing edge of the first three-UI-wide preamble pulse is then compared to that reference (average value) to obtain jitter values for display as jitter waveform, histogram of jitter, or FFT spectrum analysis of jitter. The three-UI pulse in a preamble is the most robust portion of the digital interface signal, since it is least affected by reduced bandwidth in the cable or system. Therefore, jitter measurements made with the Preamble jitter detection selection tend to be measurements of the intrinsic jitter in the transmitting device clock and are relatively unaffected by data jitter caused by reduced bandwidth. Since this derived reference clock rate is low (twice the audio sample rate), the effective jitter measurement bandwidth equals the audio sample rate when Preamble is selected.

The Stable Bits jitter detection selection derives the stable reference clock at 1/4 the actual cell (bit) rate, synchronized to the beginning transition of the preamble. The serial signal consists of 32 cells (bits) per subframe and two subframes (left and right channels) per frame. The frame rate is equal to the sample rate of the audio. Thus, there are 64 cells (bits) in a complete frame and the cell rate is 64 times the audio sample rate. The first four cells of each subframe are the preamble. The preamble always starts with a three UI (1 1/2 cell) wide pulse followed by sequences of one UI, two UI, and three UI pulses which are different among the three possible preambles. There is no cell transition time within the preamble which is common to all three preambles. The highest rate at which transitions can be guaranteed to occur regularly is at 1/4 the cell rate, which includes the beginning and end of each preamble but no transitions within the preamble. This rate is 16 times the audio sample rate, so the effective jitter measurement bandwidth is eight times the audio sample rate (384 kHz at a 48 kHz sample rate).

The All Bits jitter detection selection derives the stable reference clock at the actual cell (bit) rate. Since there are 64 cells per frame and the frame rate is the audio sample rate, the reference clock is at 64 times the sample rate and the effective jitter measurement bandwidth is 32 times the audio sample rate (1.536 MHz at a 48 kHz sample rate). Since the preamble of each sub-frame will not have transitions at every cell boundary due to its three-UI-wide pulses (violations of bi-phase coding), the DSP interpolates where transitions would have occurred if the preamble did not violate bi-phase coding.

Using the Jitter Detector is another method of measuring jitter when single readings are required (as opposed to a graphic display) for AES 3 or IEC 60958 type specifications or when measuring sample rates above 48 kHz. The Jitter Detector is visible on the Digital I/O panel's maximized view. Also, while jitter accuracy is not defined within the Intervu macros, it is specified for Jitter Measurement (via the Jitter Detector) as ± (10% + 1.0 ns), peak calibrated. consists of 32 cells (bits) per subframe and two subframes (left and right channels) per frame. The frame rate is equal to the sample rate of the audio. Thus, there are 64 cells (bits) in a complete frame and the cell rate is 64 times the audio sample rate. The first four cells of each subframe are the preamble. The preamble always starts with a three UI (1 1/2 cell) wide pulse followed by sequences of one UI, two UI, and three UI pulses which are different among the three possible preambles. There is no cell transition time within the preamble which is common to all three preambles. The highest rate at which transitions can be guaranteed to occur regularly is at 1/4 the cell rate, which includes the beginning and end of each preamble but no transitions within the preamble. This rate is 16 times the audio sample rate, so the effective jitter measurement bandwidth is eight times the audio sample rate (384 kHz at a 48 kHz sample rate).

The All Bits jitter detection selection derives the stable reference clock at the actual cell (bit) rate. Since there are 64 cells per frame and the frame rate is the audio sample rate, the reference clock is at 64 times the sample rate and the effective jitter measurement bandwidth is 32 times the audio sample rate (1.536 MHz at a 48 kHz sample rate). Since the preamble of each sub-frame will not have transitions at every cell boundary due to its three-UI-wide pulses (violations of bi-phase coding), the DSP interpolates where transitions would have occurred if the preamble did not violate bi-phase coding.

Using the Jitter Detector is another method of measuring jitter when single readings are required (as opposed to a graphic display) for AES 3 or IEC 60958 type specifications or when measuring sample rates above 48 kHz. The Jitter Detector is visible on the Digital I/O panel's maximized view. Also, while jitter accuracy is not defined within the Intervu macros, it is specified for Jitter Measurement (via the Jitter Detector) as ± (10% + 1.0 ns), peak calibrated.