Interconnect Network Modeling •Motivation: Investigate the response of a complex interconnect network to external RF interference or internal coupling between different parts of a system •On-chip interconnects on lossy substrate: capacitively and inductively coupled •Characterized with S-parameter measurements •Equivalent circuit models found by parameter-fitting Modeling a Complex Interconnect Network • Full-chip electromagnetic simulation: Too computationintensive – On-chip scale changes; too many mesh points… – Simulation can not be repeated for all possible inputs • Possible to do full wave simulation for small “unit cells”: Simple seed structures of single and coupled interconnects (Bo Yang) • Create an equivalent circuit model from EM simulation results/S-parameters • We have already developed a methodology to solve a large network composed of unit cells with random inputs while investigating chip heating (Akin Aktürk) Modeling a Complex Interconnect Network • Example Goal: Evaluate the sensitivity of different interconnect layouts to external pulses – Obtain unit cell equivalent circuits from full wave simulation/measurement – Set up a coupled network from unit cells, which emulates a certain type of interconnect network layout – Calculate the impulse responses {hi_j | i, j within chip} over time at each selected output point xi in the network, for input impulses at every possible coupling point xj – The interconnect network is a linear time invariant system: It is straightforward to calculate the output to any input distribution in space and time from the impulse responses. – Generate a random distribution of external input pulses to find the response and create a coupling map for this type of layout – Compare different layouts (i.e. different unit cell network configurations). Calculating the response to a general input from impulse responses (One-dimensional input Assume we can calculate the system’s response to an function, changing in time) impulse at point x : i (x-xi,t-t’)=(x-xi)(t-t’) hi(x,t) Take an input function applied only at one spatial point, time-dependent: f ( xi , t ) f ( x, t ) ( x xi ) f ( x, ) ( x xi ) (t )d f(x ,t) i Fi(x,t), where Fi ( x, t ) f ( xi , )hi ( x, t )d For a linear system, superposition holds: f ( x, t ) f ( xi , t ) F ( x, t ) Fi ( x, t ) i i F ( x, t ) f ( xi , )hi ( x, t )d i The discrete case Define the unit impulse: [x-xi][t-t’] [xi]= 1, x=xi 0, else hi[x,t] Input function applied only at one spatial point: f [ xi , t ] f [ x, t ] [ x xi ] f [ x, t j ] [ x xi ] [t t j ] j Fi[x,t], where f[x ,t] i Fi [ x, t ] f [ xi , t j ]hi [ x, t t j ] For a linear system, if j f [ x, t ] f [ xi , t ] F[ x, t ] Fi [ x, t ] i i F [ x, t ] f i [ xi , t j ]hi [ x, t t j ] i j The discrete case f [t ] fi [t ] F[t ] Fi [t ] F [t ] f [ xi , t j ]hi [ x, t t j ] i j The input values at discrete points in space and time can be selected randomly, depending on the characteristics of the interconnect network (coupling, etc.) and of the interference. ij : f [ xi , t j ] F [t ] ij hi [ x, t t j ] i j The Computational Advantage F [t ] ij hi [ x, t t j ] i j •Choose a spatial mesh and a time period •Calculate the impulse response over all the period to impulse inputs at possible input nodes (might be all of them) •Then we can calculate the response to a random input by only summation and time shifting •We can explore different random input distributions easily, more flexible than experimentation A simple demonstration •Simulate impulse responses at points F1..F6 to impulse inputs at points F1..F6, at time t=0 •Simulate response at point F3 to a discrete-time input given by 2 ( x x2 , t ) 3 ( x x2 , t 200ns) 3 ( x x5 , t 100ns) ( x x5 , t 400ns) A simple demonstration •Theoretically, the response at point F3 to a discrete-time input 2 ( x x2 , t ) 3 ( x x2 , t 200ns) 3 ( x x5 , t 100ns) ( x x5 , t 400ns) should be 2h3_ 2 (t ) 3h3_ 2 (t 200ns) 3h3_ 5 (t 100ns) h3_ 5 (t 400ns) •Calculate this analytically from the simulated impulse responses and compare with simulation result A simple demonstration A simple demonstration