April 5-7, 2022|Santa Clara Convention Center| Santa Clara, CA
Speaker:
Masashi Shimanouchi (Design Engineer, Intel)
Authors:
Mike Li (Fellow, Intel)
Hsinho Wu (Design Engineer, Intel)
Location: Ballroom D
Date: Thursday, April 7
Time: 2:00 pm - 2:45 pm
Track: 08. Measurement & Simulation Techniques for Analyzing Jitter, Noise & BER
Format: Technical Session
Education Level: All
Pass Type: 2-Day Pass, All Access Pass
Vault Recording: TBD
Audience Level: All
802.3 and OIF/CEI specifies the required characteristics of high speed serial link building blocks to guarantee the interoperability. Interference tolerance test is a critical compliance item for a receiver to work properly with degraded signal.
The amount of the signal degradation is calibrated with COM, and engineers must provide their own reference transmitter jitter. COM requires random jitter (RJ) and dual-Dirac deterministic jitter (ADD), which are converted by the standard-specified equations from the measured jitter Jnu and Jrms of their own signal generator.
The issue is that (RJ,ADD) converted from (Jnu/Jrm) using the current standard's method can be very inaccurate under certain conditions where its approximation is improper. In this paper, we discuss mathematically accurate solution for this conversion. Then we propose two new methods with keeping the existing framework and/or equations as much as possible.
The first method is to use a lookup table utilizing the mathematically accurate solution. There is no closed form solution for this set of non-linear equations. Using look table with existing two equations is equivalent to providing the 3rd equation to solve the problem with three unknows. The second method is to add more conditional data processing to improve approximation.
Receiver interference test requires engineers to obtain jitter (RJ,ADD), which are converted from measured jitter (Jnu/Jrm) of their own signal source using standard-specified equations. This conversion can be very inaccurate under certain conditions. We discuss mathematically accurate solution, and propose two new practical methods to improve the accuracy.