https://open-source-silicon.dev logo
#analog-design
Title
# analog-design
s

Stefan Schippers

10/12/2020, 11:19 PM
@User i will make a correction in the models, from @User i have understood that multiplicity factor
m
and
mult
must both be given. The '`m=` is the actual subcircuit multiplier (and the pfet is wrapped in a subcircuit), the '`mult=`' gets the same value as
m=
, this is only to make the 'mult' variable accessible to be used in equations inside the subcircuit. (the 'm' factor gets resolved by spice parser and no more 'visible' from within the subckt).
w

Weston Braun

10/12/2020, 11:58 PM
Is this something that is going to be changed in the model files that are distributed or do we need to specify both "m" and "mult"? I defined a value for mult and tried values equal to and different than what I had defined for "m" and did not see any differences
or was there some guidance that we had to define both and you are clarifying that that is actually not needed
t

Tim Edwards

10/13/2020, 12:00 AM
@Weston Braun: "mult" should be set equal to "m", but I don't think it shows up in any equations other than monte carlo ones.
s

Stefan Schippers

10/13/2020, 1:42 AM
Usually (if i remember correctly) there should be expressions like deltavt = random_var * coeff/sqrt(W*L*mult); multiple devices average variations reducing sigma , as do big W and big L
t

Tim Edwards

10/13/2020, 2:39 AM
@Stefan Schippers: Yes, that's exactly right.
p

Patrick Van Oosterwijck

10/13/2020, 3:32 AM
Oh thanks for that explanation. I keep bumbling from surprise to surprise. I wonder how many more ways I can be doing it wrong. 😉