Download presentation
Presentation is loading. Please wait.
1
Chapter 3b Static Noise Analysis
Cx Victim net Aggressor net Prof. Lei He Electrical Engineering Department University of California, Los Angeles URL: eda.ee.ucla.edu
2
Outline Introduction and Motivation Noise Models RC Model
J. Cong, Z. Pan and P. V. Srinivas, "Improved Crosstalk Modeling for Noise Constrained Interconnect Optimization", ASPDAC 2001 Worst case noise for RC Lauren Hui Chen, Malgorzata Marek-Sadowska: Aggressor alignment for worst-case coupling noise. ISPD 2000: 48-54 Worst case noise for RLC Jun Chen and Lei He, "Worst-Case Crosstalk Noise for Non-Switching Victims in High-speed Buses", TCAD, Volume 24, Issue 8, Aug. 2005, Pages:
3
Victim net Aggressor net
Introduction Coupling Capacitance Dominates Signal delay Crosstalk noise What is Crosstalk noise? Capacitive coupling between an aggressor net and a victim net leads to coupled noise Aggressor net: switches states; source of noise for victim net Victim net: maintains present state; affected by coupled noise from aggressor net Cx Victim net Aggressor net output
4
Noise Models RC model Worst case noise for RC Worst case noise for RLC
J. Cong, Z. Pan and P. V. Srinivas, "Improved Crosstalk Modeling for Noise Constrained Interconnect Optimization", ASPDAC 2001 Worst case noise for RC Lauren Hui Chen, Malgorzata Marek-Sadowska: Aggressor alignment for worst-case coupling noise. ISPD 2000: 48-54 Worst case noise for RLC Jun Chen and Lei He, "Worst-Case Crosstalk Noise for Non-Switching Victims in High-speed Buses", TCAD, Volume 24, Issue 8, Aug. 2005, Pages:
5
Aggressor / Victim Network
Assuming idle victim net Ls: Interconnect length before coupling Lc: Interconnect length of coupling Le: Interconnect length after coupling Aggressor has clock slew tr
6
Victim net is modeled as 2-π -RC circuits Rd: Victim drive resistance
Cx is assumed to be in middle of Lc Rise time victim / aggressor coupling capacitance
7
2- π Model Parameters Aggressor Victim
8
Analytical Solution
9
Analytical Solution part 2
s-domain output voltage Transform function H(s)
10
Analytical Solution part 3
Aggressor input signal Output voltage
11
Simplification of Closed Form Solution
Closed form solution complicated Non-intuitive Noise peak amplitude, noise width? Dominant-pole approximation method
12
Dominant-Pole Simplification
RC delay from upstream resistance of coupling element Elmore delay of victim net
13
Intuition of Dominant Pole Simplification
vout rises until tr and decays after vmax evaluated at tr
14
Extension to RC Trees Similar to previous model with addition of lumped capacitances Extended to a victim net in general RC tree structure
15
Results Average errors of 4% comparing to HSPICE in peak noise and noise width. Devgan model 589% Vittal model 9% 95% of nets have errors less than 10%
16
Spice Comparison peak noise noise width
17
Effect of Aggressor Location
As aggressor is moved close to receiver, peak noise is increased Ls varies from 0 to 1mm Lc has length of 1mm Le varies from 1mm to 0
18
Optimization Rules Rule 1: If RsC1 < ReCL
Sizing up victim driver will reduce peak noise If RsC1 > ReCL and tr << tv Driver sizing will not reduce peak noise Rule 2: Noise-sensitive victims should avoid near-receiver coupling
19
Optimization Rules part 2
Preferred position for shield insertion is near a noise sensitive receiver Rule 4: Wire spacing is an effective way to reduce noise Rule 5: Noise amplitude-width product has lower bound And upper bound
20
Noise Models 2-Pi model Worst case noise for RC
J. Cong, Z. Pan and P. V. Srinivas, "Improved Crosstalk Modeling for Noise Constrained Interconnect Optimization", ASPDAC 2001 Worst case noise for RC Lauren Hui Chen, Malgorzata Marek-Sadowska: Aggressor alignment for worst-case coupling noise. ISPD 2000: 48-54 Worst case noise for RLC Jun Chen and Lei He, "Worst-Case Crosstalk Noise for Non-Switching Victims in High-speed Buses", TCAD, Volume 24, Issue 8, Aug. 2005, Pages:
21
Worst Case Noise Model Consider multiple-aggressors situation:
Each aggressor (A1, …, A5) has its switching signal. Each switching aggressor will result in a coupling noise on victim at variable arrival times.
22
Worst Case Noise Model To consider Worst Case Noise (WCN):
Make alignment of aggressor inputs (change arrival time) The coupling noise at victim output can occur at the same time. Aggressor Alignment Problem Formulation: Find the relative relationships among arrival times for all aggressor inputs such that all individual peak noises are aligned, assuming all the other conditions are fixed.
23
WCN Superposition Consider two aggressors (V1 and V2) case
N1: when V1 is switching, V2 is quiet; N2: when V2 is switching, V1 is quiet; Individual noise waveforms
24
WCN Superposition To consider WCN, the aggressor alignment is performed: Change the arrival time of V2 Two noise signals can occur at the same time
25
WCN Analysis strategies
Four WCN analysis strategies based on aggressor alignment Explicit Aggressor Alignment (AS: Aligned switching) Noise output is obtained by aligning switching of all aggressors. The largest amplitude is WCN. No Aggressor Alignment (SS: simultaneous switching) Simultaneous switching of all aggressors. Implicit Aggressor Alignment (SP: Superposition) Each noise output is obtained with only one aggressor switching; Total peak noise is the summation over all individual peak noise. Extension of Implicit Aggressor Alignment “back-annotates”: use output noise to determine the aggressor input skews, and estimate the coupling stage again.
26
Peak Noise Comparisons with no timing constraints
Experiment Peak Noise Comparisons with no timing constraints WCN analysis can provide higher accuracy than superposition does.
27
Peak Noise Comparisons when timing constraints are given
Experiment Peak Noise Comparisons when timing constraints are given Similarly, WCN analysis can provide higher accuracy than superposition does.
28
Noise Models 2-Pi model Worst case noise for RC
J. Cong, Z. Pan and P. V. Srinivas, "Improved Crosstalk Modeling for Noise Constrained Interconnect Optimization", ASPDAC 2001 Worst case noise for RC Lauren Hui Chen, Malgorzata Marek-Sadowska: Aggressor alignment for worst-case coupling noise. ISPD 2000: 48-54 Worst case noise for RLC Jun Chen and Lei He, "Worst-Case Crosstalk Noise for Non-Switching Victims in High-speed Buses", TCAD, Volume 24, Issue 8, Aug. 2005, Pages:
29
Worst Case Noise (WCN) for RLC tree
Problem Formulation: Given a non-switching victim and multiple aggressors in a pre-routed interconnect structure Object: find switching patterns and switching times for all aggressors such that the noise in the victim has maximal amplitude. Recall basic WCN analyses for RC model: SS: Simultaneous switching SP: Superposition AS: Aligned switching How to extend WCN analysis to the RCL model?
30
Shielding: Vdd s1 s2 s3 s4 Gnd WCN under the RCL model
Dedicated shields can reduce crosstalk noise. Assume there are shields at both edges of the bus structure. Vdd s1 s2 s3 s4 Gnd
31
Switching Pattern WCN under the RCL model
Waveform can have resonance due to inductance under RCL model Resonance leads to multiple noise peaks with opposite polarities. WCN may happen when aggressors switch in the same or different direction. V – quiet victim q – q quiet wire a - aggressor S - shield
32
WCN analysis under RLC model
Extension to Existing Algorithm for RC Simultaneous Switching (SS): All aggressors switch simultaneously in the same direction WCN is the maximum noise on the victim Superposition (SP) Find maximum noise peak for each aggressor when only this aggressor switches. WCN is the summation of amplitudes of all such peaks.
33
WCN analysis under RLC model
AS (Aligned Switching) Find individual noise with only one aggressor switching; Switch multiple aggressors to find the maximum noise PP alignment: align the maximum positive peaks of individual noises all aggressors switch in the same direction NN alignment: align the maximum negative peaks of individual noises PN alignment: align the peaks of maximum amplitude Aggressors have switching directions that all the aligned peaks have the same polarity. WCN is the maximum noise among the above simulations.
34
Experiment Noises on a quiet victm from different algorithms for an aligned RLC bus structure Accurate results SA+GA select larger noise from annealing algorithm (SA) and genetic algorithm (GA) as the accurate WCN. SS + ASWCN is approximated by the larger one between the results obtained by SS and AS. SS + AS gives results very close to SA + GA WCN under the RC model severely underestimates the noise in most cases, especially for strong drivers and larger spacing.
35
Experiment Noises on a noisy victm from different algorithms for an aligned RLC bus structure Similarly, SS + AS gives results very close to SA + GA. SP severely underestimates WCN, witha maximum underestimation of 39.93% and an average underestimation of 20.53%.
36
Noise Models Noise reduction 2-Pi model Worst case noise for RC
J. Cong, Z. Pan and P. V. Srinivas, "Improved Crosstalk Modeling for Noise Constrained Interconnect Optimization", ASPDAC 2001 Worst case noise for RC Lauren Hui Chen, Malgorzata Marek-Sadowska: Aggressor alignment for worst-case coupling noise. ISPD 2000: 48-54 Worst case noise for RLC Jun Chen and Lei He, "Worst-Case Crosstalk Noise for Non-Switching Victims in High-speed Buses", TCAD, Volume 24, Issue 8, Aug. 2005, Pages: Noise reduction L. He and K. M. Lepak, "Simultaneous shield insertion and net ordering (SINO) for capacitive and inductive coupling minimization", ISPD 2000
37
Capacitive Coupling Noise
Coupling noise from two adjacent aggressors to the middle victim wire with 2x min. spacing. Rise time is 10% of project clock period. Capacitive coupling depends strongly on both spatial and temporal relations!
38
Layout to reduce impact of Cx
Vdd Gnd s1 s2 s3 s4 Coplanar parallel interconnect structures with pre-routed Vdd and Gnd Noise avoidance technique: Shield insertion Shield is a wire directly connected to Vdd or Gnd Vdd s1 s2 G s3 s4 Gnd
39
error occurs Signal levels (V) aggressor VIH victim1 victim2 time
Timing Sensitivity Two nets are considered sensitive if a switching event on signal s1 happens during a sample time window for s2 time aggressor Signal levels (V) victim1 VIH Sampling window error occurs victim2 no error occurs
40
Layout to reduce impact of Cx
Vdd Gnd s1 s2 s3 s4 Coplanar parallel interconnect structures with pre-routed Vdd and Gnd Noise avoidance techniques: Net ordering (track assignment / net placement) Vdd s4 s1 s3 s2 Gnd
41
Characteristics of Coupling Capacitance
Coupling capacitance is highly sensitive to spacing Proper wire sizing and spacing may limit the impact of Cx by changing the ratio Cx/Ctotal E1 spacing Spacing (nm)
42
Relation between Delay and Noise
T_max = T * ln (1/0.5-v) / ln 2 T_min = T * ln (1/(o.5+v) / ln 2 Typical values V T_max/T T_min/T AGG AOUT VOUT VIC VOUT w/o XTalk delay
43
Noise estimation and filtering
Rule of thumb: Cx/C < threshold Devgan, ICCAD’97 V < (Rv + Rint / 2) * Cx / (1.25 Tr) Tr: rising time for the aggressor Vittal et al, TCAD’99 (more accurate) V = (Rv + Rint / 2) * Cx / {0.63Tr + Ra (Ca + Cx) + Rv (Cv + Cx) + Rint * Cx} To reduce Cx impact Increase the driver size of victim Decrease the driver size of aggressor Or buffering Need a global device size solution coupled with Time Analysis
44
Capacitive crosstalk is localized
Mini-Summary Capacitive crosstalk is localized Capacitive crosstalk affects both delay and signal integrity Capacitive crosstalk can be minimized by Spacing (and wire sizing) Device sizing Net ordering Shielding Buffering I will first briefly give the background and overview of my dissertation, then cover more details for an important component of the dissertation: the LR-base STIS optimization. Here LR refers to local refinement, and STIS refers to simultaneous transistor and interconnect sizing. Finally, we draw conclusions and discuss future works.
45
Candidates for On-Chip Inductance
Wide clock trees Skews are different under RLC and RC models Neighboring signals are disturbed due to large clock di/dt noise Fast edge rate (~100ps) buses RC model under-estimates crosstalk P/G grids (and C4 bumps) di/dt noise might overweight IR drop I will first briefly give the background and overview of my dissertation, then cover more details for an important component of the dissertation: the LR-base STIS optimization. Here LR refers to local refinement, and STIS refers to simultaneous transistor and interconnect sizing. Finally, we draw conclusions and discuss future works.
46
Inductance Minimization
Reference plane wiring layers sandwiched between power planes GND plane VDD plane
47
Inductance Minimization
Coplanar shields Bus signals VDD shield GND shield
48
Characteristics of Coupling in 18-Bit Bus
# of Shields Noise (% of Vdd) 0 (a) 0.71V (55%) 2 (b) 0.38V (29%) 5 (c) 0.17V (13%) (a) (b) (c) Lx coupling between non-adjacent nets is non-trivial Shielding is effective to reduce Lx coupling
49
Figure of Merit of Inductive Coupling
Inductive coupling coefficient defined as A formula-based Keff model has been developed High fidelity between formula and noise voltage [He-Xu, 2000]
50
Illustration of Keff Computation [XuHe,2000]
Keff(i,j) = (f(i) + g(j)) / 2 f(i) = (Ni – gl)/(Nj – gl); g(j) = (gr – Nj)/(gr-Ni)
51
Inductance Minimization
Staggered inverters/buffers Differential signals Nets with opposite switching signals can be placed adjacent to each other Decrease Lx noise at the cost of a higher Cx noise Mutual capacitance polarities
52
Mini-Summary Inductive crosstalk is globalized
Inductive crosstalk affects both delay and signal integrity Inductive crosstalk is not sensitive to Spacing (and wire sizing) Net ordering Inductive crosstalk can be minimized by Shielding Buffering Ground plane Differential signal Signal termination
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.