MODELING AND BUDGETING OF TIMING JITTER AND NOISE - Advanced Signal Integrity for High-Speed Digital Design - page 590

120

100

80

60

40

20

0

PDA

− 20

− 40

− 60

− 80

− 100

− 120

0 10 20 30 40 50 60 70 80 90 100

Time (ps)

Figure 13-27 Comparison of eyes for a 200-bit random sequence and the worst case
from peak distortion analysis.

impacts of crosstalk on the data eye to make sure that it is comprehended in the
system jitter and noise margin estimates. Fortunately, the superposition principle
makes it possible to extend the PDA method to include crosstalk using the same
approach that we employed for a single signal.
To illustrate the concept, we return to the coupled system from Example 4-2.
Shown in Figure 13-28, the system contains a 0.2794-m-long coupled pair that is
terminated with 65 at each end. The lossless, single-ended signals are driven
by a 1-V source with a 100-ps rise time and 1-Gb/s switching rate. The signals
are detected by a high-speed differential receiver that has one end connected to
a reference voltage of 0.25 V. To perform peak distortion analysis, we again
simulate the system pulse response, which is shown in Figure 13-29a. When
including crosstalk, we apply the active pulse to the line of interest (the “aggres-
sor”) and capture the pulse response at the receiver for all signals in the coupled
system.
We start by using PDA to calculate the worst-case signal waveforms that
account for ISI from the pulse response on the aggressor line. To incorporate the
effect of crosstalk, we use the response on the victim line in the same manner
as we did for the ISI calculation. That is, we conditionally add the crosstalk
at multiples of the unit interval to the worst-case signals that include ISI at
the current sample point. For a worst-case 1, the samples that contain negative
crosstalk are added to the pulse response. For the worst-case zero calculation, the
samples containing positive crosstalk are added to the steady-state low response.