shikhanshu
Newbie level 2
hspice gauss
hello,
i want to see the effect of random parameter variation (parameters like width, tox etc) on gate delay/power/area etc... but this randomness is correlated spatially... meaning, i want closely placed transistors to vary similarly...
i know how to declare gaussian variables in hspice...
.param width_param = gauss(mean, variation, sigma) et al...
and then i will run a monte carlo for some iterations and in each iteration hspice itself will generate a new width_param value that i can use for a nmos/pmos device...
HOWEVER! i dont know how to declare correlated variables... say, i want to declare 4 such gaussian variables (say i have 4 transistors in my gate and i want the width of each as a separate gaussian variable) ...
but i also want all these 4 gaussian variables to be correlated using a 4x4 correlation matrix... is that even possible in hspice? if not, any suggestions on how can i go about doing this?
thank you.
hello,
i want to see the effect of random parameter variation (parameters like width, tox etc) on gate delay/power/area etc... but this randomness is correlated spatially... meaning, i want closely placed transistors to vary similarly...
i know how to declare gaussian variables in hspice...
.param width_param = gauss(mean, variation, sigma) et al...
and then i will run a monte carlo for some iterations and in each iteration hspice itself will generate a new width_param value that i can use for a nmos/pmos device...
HOWEVER! i dont know how to declare correlated variables... say, i want to declare 4 such gaussian variables (say i have 4 transistors in my gate and i want the width of each as a separate gaussian variable) ...
but i also want all these 4 gaussian variables to be correlated using a 4x4 correlation matrix... is that even possible in hspice? if not, any suggestions on how can i go about doing this?
thank you.