LeCroy SDA, Operator'S Manual

Page: 386 / 397

The Relative Impact of Noise on Jitter (Quadratic Addition of Noise)
The “quadrature addition” used to “add” the noise components is really only precise for Gaussian
distributions of variations (even though this works well for many combinations of incoherent
phenomena).
What’s important in this application is that (at least for current instruments) the instrument’s
contribution to vertical noise is almost always Gaussian in nature. Furthermore, this can be
confirmed by in-situ calibration procedures.
How the Measurement System Noise Is Subtracted from the Jitter Measurement
The timing noise can be very complex, exhibiting periodic jitter ( Pj ) and, for data streams, ISI
induced data dependent jitter DDj . Other forms of bounded uncorrelated jitter ( BUj ) can
contribute, as well, to make the jitter distribution shape quite non-Gaussian.
If, however (as is usually the case), the jitter analysis breaks down the distribution characteristic
into
Rj and Dj , the systematic effects are lumped into the Dj part. The Rj part of this jitter behaves
as a Gaussian, and the vertical noise contribution (from the instrument) can be compensated as
follows:
So, if the dv/dt can be well estimated by the instrument, and if the instrument’s contribution to
vertical noise is known (and incoherent with the signal, which is nearly always the case), this
compensated Rj figure can be reported.
As such, the LeCroy SDA software has been equipped to perform this compensation, and applies
it to the
Rj figure and not the Dj figure; however, the corresponding effect on Tj is reported as
well.
Per model noise figures are supplied as a built-in database for each SDA oscilloscope, and a
procedure for updating the values for a specific oscilloscope channel and probe are provided.
Q-scale Theory
Introduction
Jitter is an important aspect of signal integrity for both optical and electrical serial data streams
(and clocks). The SDA (serial data analysis) software is designed to measure the jitter and its
components: random jitter (Rj), deterministic jitter (Dj), data dependent jitter (DDj) duty cycle
distortion (DCD), and periodic jitter (Pj). The SDA uses a powerful method called “Normalized Q-
scale Analysis” to estimate/measure the random and bounded, uncorrelated jitter components.
The following section presents the technical background underlying this method.
386
SDA-OM-E Rev H