The PLL Clock (Recovered Clock) is to be used as the external trigger source. Peak-to-peak jitter amplitude should be measured at the 50% crossing point of the differential clock signal.

The specification further recommends that the random jitter at 10-9 pixel error rate should be +6 sigma limit points of the distribution.

However, jitter might include random as well as deterministic (systematic) components and hence care must be taken to eliminate the systematic components as otherwise these would lead to an over-estimated value of sigma.

Performing Jitter Measurement

DVI Test and Measurement guide suggests the following procedures for performing the jitter measurement.

Method I:
The differential clock signal must be accumulated for at least 100,000 acquisitions and jitter measurement should be performed as mentioned above (by removing the systematic jitter component).

Method II:
Accumulate at least 1,000,000 acquisitions and read the peak-peak jitter amount using the scope histogram measurements (peak-to-peak jitter) as shown in Fig. 10.

Fig. 10

However, both the above-defined methods do not make a mention of the inherent scope trigger jitter that can become a significant contributor to the jitter value.

TDSJIT3 Jitter and Timing analysis package can be used to separate the Random and Deterministic components of jitter, view the jitter trends, analyze Jitter Spectrums and view the bathtub curve and eye opening at different bit error rates.

Fig. 11 - The above screen shot shows TDSJIT3 separating the random and deterministic components of jitter.

Fig. 12 - TDSJIT3 also shows bathtub curves that provide information regarding the eye opening at different bit error rates.

Fig. 13 - The screen shot is the magnitude spectrum of the TIE jitter measurement.

For more information view the TDSJIT3 - Rj/Dj Application Note