•  
• EDN Europe MAGAZINE
•  
• Free SUBSCRIPTION
• Subscription Renewal
• View Media Kit

• EDN
• EDN Australia
• EDN Asia
• EDN China
• EDN Japan

SIGNAL INTEGRITY : Working with jitter statistics

Iregularly receive emails discussing the subtleties of low-voltage differential signaling, pre-emphasis and equalisation. I apologise if I’ve yet to respond to you personally, but as several correspondents pointed out, the illustration in the previous article in this series (Reference 1) did show both amplitude and phase errors. Moreover, many of you rightly pointed out to me that a significant side effect of edge restoration is the possibility of generating timing errors, otherwise known as jitter.

While many books and white papers both define and describe the measurement of jitter, one question often remains: Exactly what is the jitter budget for a particular design? My answer is that this is often a matter of risk assessment where, in the absence of any benchmark or compliance requirement, a designer will estimate a statistical threshold of jitter that his or her digital design will tolerate. In other words, if a jitter measurement has a statistical basis, and those statistics in turn contain an element of uncertainty, the designer has no alternative but to take a calculated risk. To demystify jitter, this and the next article will explain how to interpret and apply jitter statistics, starting by reviewing the basics of jitter definition and measurement.

Jitter is often defined as the shortterm variations of a digital signal’s transitions or edges as it switches between logic states, where the transitions deviate from their ideal positions in time. Figure 1 is an oscillogram of a signal generator output waveform with added jitter, in which the capture and overlay of several cycles clearly shows the jitter at the signal transitions. The highperformance oscilloscope is a primary tool for visualising and measuring jitter where built-in clock recovery, eye diagrams, compliance masks, bathtub curves and statistical models provide quantitative signal analysis.

The eye diagram has become the definitive tool for validation and compliance testing of digitally transmitted signals. The eye diagram comprises superimposed waveform traces from many successive UIs (unit intervals). Eye diagrams display serial data with respect to a clock that hardware or software tools recover from the data signal using either hardware or software tools. In the example shown in Figure 2, a hardware- based “golden phase locked loop” recovered the clock. The diagram displays all possible transitions, edges, positive- going and negative-going, and both data states in a single window. Using some imagination, the image resembles an eye.

In an ideal world, each new trace would line up perfectly on top of those that came before it. In the real world, however, signal-integrity factors such as noise and jitter cause the composite trace to “blur” as successive cycles are accumulated.

The blue regions in Figure 2 have special significance. They are the violation zones used as mask boundaries during compliance testing. The blue polygon in the centre defines the area in which the eye is widest. This encompasses the range of safe decision points for extracting the data content from the received signal. The upper and lower blue bars define the signal’s amplitude limits. If a signal penetrates the mask, it constitutes a “mask hit” that may cause the compliance test to fail. Mask hits have to be quantified, as some standards tolerate a small number of mask hits.

Jitter measurements have been a topic of extensive discussion in industry SIGs (special interest groups) in recent years. Jitter is so important that specialised analysis tools now help designers penetrate this difficult problem. TIE (time-interval error) is the basis for many jitter measurements. TIE is the difference between the recovered clock, the jitter timing reference, and the actual waveform edge. Performing histogram and spectrum analysis on a TIE waveform provides the basis for advanced statistical jitter measurements. Having revisited these fundamental concepts, in the next article I will address interpreting and applying jitter statistics.