Monte Carlo Simulation for MOS mismatch

Dear All:



I'm working on a mos mismatch investigation for a circuit at cadence.
But I can't get how to do mismatch set for chennel length or width. In
statistics file, there's only place for parameters in model. If I have
size mismatch, how can I put they in statistics file? By the way, is
there anyone know how to deal with correlate factor between W and L at
same device?

Thank you very much.


Huang

 

[Andrew Beckett]

On 17 Sep 2005 03:49:59 -0700, hb_huang@hotmail.com wrote:

Dear All:



I'm working on a mos mismatch investigation for a circuit at cadence.
But I can't get how to do mismatch set for chennel length or width. In
statistics file, there's only place for parameters in model. If I have
size mismatch, how can I put they in statistics file? By the way, is
there anyone know how to deal with correlate factor between W and L at
same device?

Thank you very much.


Huang

Start by reading http://www.designers-guide.org/Modeling/montecarlo.pdf and then
seeing if you have any other questions.

Also search this forum on google, and look on the forum on the designer's guide
site.

Regards,

Andrew.

 

Hi Andrew:


Thanks for reply. I've gone through with Don's paper from
designers-guide and get some ideas. For some parameters from BSIM model
or other model (sub cuicuit), Don's method worked great, like: Vth0,
tox, or rshsp. However, one of my main questions is still exist. It's
main regarding to geometrical mismatch issue. I don't know if there
is/are parameters at BSIM MOS model for width and length scaling
factor. At Don's paper, he introduced such factors XRSPbfac and
XRSPafac by including a inline sub circuit of resistor. By this mean,
do I need to do same thing for MOS?

Also, generally I would use Vth0, tox for MOS process and mismatch
variation, if it's enough for normal purpose or not?


Regards


Huang

 

[okguy]

Take care of the parameter : m
You cannot acces it in the models.
Your mismatch varies with the square root of W.
But W varies with m.
Let's try a transistor 9/1 and another 3*3/1.

OkGuy

 

Very interesting, OkyGuy. Would you mind make it more clearly? If I
can't access to m, how can I use it as a parameter in statistics.scs
file? As a mater of fact, MC simulator at Cadence could not recognize
my m at statistics.scs file. Thanks.



Regards

Vincent

 

[okguy]

You can find more info from that post :
http://www.talkaboutcad.com/group/comp.cad.cadence/messages/17475.html