Browse Prior Art Database

Process of dealing with DC noise in a width-based noise rejection curve

IP.com Disclosure Number: IPCOM000166832D
Original Publication Date: 2008-Jan-24
Included in the Prior Art Database: 2008-Jan-24
Document File: 3 page(s) / 93K

Publishing Venue

IBM

Abstract

Traditionally, noise analysis tools have used a width-based noise rejection curve to determine if a noise event is to be considered a failure. The total noise peak and width are determined by summing the various contributing noise sources. Noise sources with disparate widths or infinite widths (DC noise) do not model well within this framework using the usual rules of computing total peak and width. Our enhancement is based on the empirical observation that, in the family of curves, the AC portion of the noise curve scales with the amount of DC noise. Instead of creating a family of curves representing the failing noise or treating noise pulses with large widths as DC noise, we propose using the measurement and storage of only a single curve, with the other curves able to be synthesized from the single measured curve with the formula provided.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 54% of the total text.

Page 1 of 3

Process of dealing with DC noise in a width -based noise rejection curve

Disclosed is a method of dealing with DC noise in a width-based noise rejection curve. Traditionally, noise analysis tools have used a width-based noise rejection curve to represent the functional tolerance of a circuit to noise. These noise rejection curves, as shown below, allow a program to evaluate a noise pulse's attributes, namely its pulse width and pulse amplitude, and decide if that noise pulse is likely to cause a functional failure in the circuit under test. When more than one noise pulse is involved, the total noise peak and width are determined by summing the various contributing noise sources

.

Fail

Noise pulse width (sec)

A problem with this method arises when noise sources with disparate widths or infinite widths (DC noise) are analyzed, as they do not model well within this framework using the usual rules of computing total peak and width.

v

w

i

V is total peak noise, W is total noise width,
vi is peak noise from aggressor i, wi is noise width of aggressor i

Noise 1

Noise 2

An obvious solution appeared to be to treat noise with large widths as DC noise, and to create a family of curves representing the failing noise-the DC noise would select the appropriate curve against which the AC noise is measured.

Noise

Peak
(V)

Pass

V==

  i i

i v

   W v

 ∑ ∑

Total noise

1

Page 2 of 3

Noise Peak
(V)

Lower DC Noise

Higher DC Noise

Noise pulse width (sec)

Problems with the obvious solution include the need for time-consuming analysis of each receiving gate for each DC noise value, as well as the large amounts of storage to save the set of curves of each receiving gate.

The proposed novel enhancement is based on the empirical observation that, in the family of curves, the AC portion of the noise curve scales with the amount of DC noise.

( )

( )

  dc V Vdc
w
K
w
V

  dc V Vdc
w
K
w
V

    2 max
)
(

 ) (
2

 ) (
1

=

    1 max
)
(

=

K(w), Vdcmax independent of DC noise
V1, V2 are AC noise margins given a width V1dc, V2dc are DC noises from source 1 & 2

Noise Peak
(V)

V1

V2

Vdcmax-V1dc

Vdcmax-V2dc

w

Noise pulse width (sec)

By taking advantage of the observed behavior with respect to DC noise, the measurement and storage of only a single curve is needed, the...