Jitter Solutions for Digital Circuits

Overview

With new high-speed I/O and bus standards increasing the data rates, the measurement of jitter is rapidly becoming a necessity for ensuring error free digital communication. Keysight Technologies, Inc. provides a wide range of solutions for the prediction, characterization, and testing of jitter.

Different market segments from telecommunications to high-speed I/O connections for data communications use the term jitter (see Figure 1). In telecommunications and enterprise arenas, jitter specifications and measurements are well documented through standards bodies, so the measurements to make are well known. In the high-speed I/O arena, many new bus standards are being introduced with little commonality in specifying and measuring jitter. This forces the designer and test engineer to really understand the different jitter measurement viewpoints and what measurement techniques will be best for them.

This brochureconcentrates on the jitter measurements required by many of today’s high-speed bus and interconnect standards — InfiniBand, PCI Express, USB, Serial ATA, etc. For information on telecommunications and enterprise jitter solutions, see the Keysight brochure “Need to test jitter?” (5988-7051EN).

Keysight provides a wide range of jitter measurement solutions (see Figure 2). The real time oscilloscope, the Keysight 54850 Infiniium with the E2681A jitter analysis software, provides the most versatile view of jitter. The sampling oscilloscope, Keysight 86100B Infiniium DCA, can be a more economical solution and is the critical tool when data rates exceed 3.2 Gb/s. When you need bit error ratio (BER) measurements, the Keysight 81250 ParBERT, 71612C error performance analyzer and 86130A BERT, construct the most precise bathtub curves and are the ultimate proof of performance. When a precision, low jitter stimulus is required, the Keysight 81133A pulse/ pattern generator is ideal.

What is jitter?

Jitter is defined as the misalignment of the significant edges in a sequence of data bits from their ideal positions. Misalignments can result in data errors. Tracking these errors over an extended period of time determines the system stability. Jitter can be due to dete ministic and random phenomena, also referred to as systematic and nonsystematic respectively.

From a general standpoint, jitter characterization involves a statistical measurement of the relative position variance of clock or data edges. Serial data communication eliminates the physical and bandwidth limitations of clock and data bus transmission by embedding the clock in the transmitted data. Since the data clock is not transmitted separately, the problem of maintaining the clock and data

temporal alignment is eliminated in the data stream. However, other issues become important, such as the minimization of jitter in data transmission, faithful clock extraction from the serial data, and network timing. These problems are manifest because of random and systematic effects.

Random Jitter

Random Jitter (RJ) is Gaussian in nature and typically results from thermal noise, shot noise, etc. In that RJ is unbounded, it is often characterized through an RMS or mean value. This allows one to assess the probability that the jitter will fall both inside and outside a specified range. When an RMS value is measured,

the peak-to-peak RJ value can be estimated through multiplication by a factor dependent upon the desired bit error ratio (BER) probability (see Figure 3). Note that these multipliers are valid for purely random jitter. When DJ is present, the BER margin is reduced by the DJ value. A significant amount of data is required to yield a statistically accurate, high confidence characterization of RJ. As seemingly small amounts of RMS RJ correspond to large peak-to-peak values, precisely assessing the full impact of the jitter is achieved by directly measuring the probability of RJ at the desired low BER’s.

Figure 3. Converting RMS to Peak-to-Peak for specific BER’s.