Monte carlo model for matching

[tritue truong]

Hi all,

I use monte carlo analysis but I have a hard time trying to model the
mismatch

We can modelize the mismatch using "statistics" block with "process" and
"mismatch". However, the mismatch factor of a transistor/resistor/ capacitor
is area dependant.
Any body has any idea how to do it?

Thank you

 

[Andrew Beckett]

If I remember rightly, this is covered in some detail in
http://www.designers-guide.com/Modeling/montecarlo.pdf

The simple way of describing how to do it is to have a normalised parameter
which gives you a gaussian distribution (i.e. mean of 0, sd of 1), and then you
can scale that distribution dependent on the area of the device, and add a
constant value to it (something like that). So the random number generation is
done independently of area, but you can then scale that in the (inline) subckt
itself.

Regards,

Andrew.

On Tue, 7 Sep 2004 08:48:27 -0400, "tritue truong" <tritue@comport-data.com>
wrote:

Hi all,

I use monte carlo analysis but I have a hard time trying to model the
mismatch

We can modelize the mismatch using "statistics" block with "process" and
"mismatch". However, the mismatch factor of a transistor/resistor/ capacitor
is area dependant.
Any body has any idea how to do it?

Thank you

--
Andrew Beckett
Senior Technical Leader
Custom IC Solutions
Cadence Design Systems Ltd

 

[fogh]

tritue truong wrote:

Hi all,

I use monte carlo analysis but I have a hard time trying to model the
mismatch

We can modelize the mismatch using "statistics" block with "process" and
"mismatch". However, the mismatch factor of a transistor/resistor/ capacitor
is area dependant.
Any body has any idea how to do it?

Thank you


If I remeber well, you should see the spread decrease with square root of device area. When you have the luck to have access to extensive chracterisation data, you could also look at the effect of aspect ratio.