[Design View / Design Solution]
Characterizing High-Speed Serial Communications Links Requires Some Analog Savvy
A six-step process helps measure, identify, and eliminate clock and data jitter on those blazing serial signals.
STEP IV: Separate Jitter Types And Components Keeping jitter out of the system requires that you be able to separate the random and deterministic jitter components. For that purpose, the technique described in this step helps with debugging and design verification as well as characterization of the system links. We now analyze some of the histograms collected in the previous sections.
Histogram Plot
The total jitter histogram is a good first look at the analysis of jitter. As mentioned in a previous section, random jitter (Rj) is assumed to have a Gaussian (normal) distribution for the purpose of modeling. That means that its probability density function is described by the well known bell curve. The time interval error (TIE) histograms associated with our PRBS-23 data are shown in Figure 11. Note that the total jitter histogram can also be multi-modal.
The histogram of Figure 11a is not necessarily okay, but that of Figure 11b definitely points to some poor design issues. As shown in Figure 10, it is easy to see that a bi-modal histogram has something to do with the rising and falling edges not being aligned in the middle (some systemic problem is messing up the histogram and making it non-Gaussian). A bi-modal histogram usually indicates significant amounts of deterministic jitter.
When both Dj and Rj components are present, the jitter histogram is generally broadened, and no longer resembles a Gaussian distribution. In that case, the difference between the left and right peak values represents Dj, and results from a crossing point that is a bit higher than it should be. This condition can be associated with DCD jitter due to a cross-talking signal with a given period. That’s why it is important that designers analyze the histograms as complementary insights to eye diagrams.
Bathtub Plot
Like the histogram, the bathtub plot offers a powerful way to look at jitter and analyze its timing. By plotting bit error rate as a function of sampling position within the bit interval, the bathtub plot represents eye opening versus bit error rate (Figure 12). (Operation at an expected maximum error rate of 10–12 has become a defacto requirement in many serial standards.) As can be observed in the figure, deterministic jitter forms the almost flat horizontal portion of the bathtub curve (gold region), while the slope portion (blue region) is due to random jitter. You can also see that that the following equation applies:
Jitter Eye Opening + Total Jitter = 1UI.
The measurement of a jitter histogram, or bathtub curve, or both, is a primary step informing the SI engineer of jitter in the system. Neither measurement, however, tells him about the individual sources of the jitter components. In the next step, we attempt to identify the root cause(s) of Dj by separating it into its components.
STEP V: Diagnose The Root Cause Of Jitter We now analyze jitter in the frequency domain, which reveals Dj components (Pj, ISI, DCD, etc.) as distinct single-frequency spurs (line spectra) that can easily be visualized to analyze their sources. These frequency domain views can include the phase noise plot, the jitter spectrum plots, or a Fast Fourier Transform (FFT) of the jitter trend.
Jitter Spectrum Of Data TIE Plot
Several techniques are available for measuring jitter on a single waveform. One such technique is to examine the spectrum of the time interval error (TIE). TIE is the timing deviations of digital-data transitions from their ideal (jitter-free) locations. In short, the TIE measures how far each active edge of the clock varies from its ideal position. TIE is important because it shows over time the cumulative effect that is produced by even a small amount of jitter[2].
Please refresh the page if you have trouble reading this text.
Search Electronic Design
Email Newsletter
Sponsored By:
Electronic Design UPDATE provides readers with late-breaking news, opinions from industry experts, and timely technology stories. It's a unique opportunity to get your product message in front of engineers, engineering managers, and corporate managers while they're reading about critical information online.
Reprints
