Browse Prior Art Database

Long and Lossy Transmission Lines for High-Speed I/O and Other Circuits

IP.com Disclosure Number: IPCOM000077222D
Publication Date: 2005-Feb-25
Document File: 2 page(s) / 26K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a method for an algorithm to model electrically long transmission lines to correct underestimated losses.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 53% of the total text.

Long and Lossy Transmission Lines for High-Speed I/O and Other Circuits

Disclosed is a method for an algorithm to model electrically long transmission lines to correct underestimated losses.

Background

There are modeling and simulation issues for off-chip interconnections for data rates faster than 2 GT/s. Today’s PCI and CSI designs require a data rate of 6-10 GT/s; by year 2005, this number will increase to 30+ GT/s. 

Currently, long T-lines are broken manually into a number of short-lines in the SPICE circuit deck; the broken lines’ time-domain response is correlated using a sinusoidal source against the corresponding frequency domain reference (see Figure 1). This process is repeated until an “appropriate” number of the line splits is determined. This is a tedious and laborious procedure. Furthermore, during the simulation stage some commercial tools begin to exhibit unpredictable instability issues, causing severe phase or delay errors.

General Description

The disclosed method address the inaccuracy issues associated with time domain simulations, when long multi-conductor transmission lines (MTL) are approximately modeled and represented in a frequency domain. As observed in applications, faster data rates create electrically longer lines, causing larger errors with the existing model approach. 

To solve the problem, the disclosed method uses the virtual node placement on the long MTL, either uniformly or non-uniformly (see Figure 2) for line segmentation. Each segment between the two adjacent nodes along the wave propagation direction is treated as an effective “short” MTL. The number of segments placed along the MTL is determined heuristically, in terms of the data rate or signal rise/fall...