Subhajit Sen
Guest
Sat Nov 22, 2003 2:35 pm
I am trying to do distortion analysis of a circuit
using a MOSFET in linear region. If I use a
BSIM3V3 model the results are incorrect because
of discontinuity in the high-order
derivatives of Id vs. vds characteristic around
Vds=0. This is a known problem. (see for example,
paper by Y. Tsividis, K. Suyama, IEEE Jl. Solid-State
Circuits, March 1994).
My question is: has anybody has been able to overcome
this problem? I am aware of EKV model but do not
have access to a practical model derived from
characterizations that I could use.
Thanks,
Subhajit Sen
(PS: Pls. also reply to subhajits_at_india.cirrus.com).
Steve Hamm
Guest
Sat Nov 22, 2003 7:19 pm
On 22 Nov 2003 05:35:58 -0800, sub_sen_at_hotmail.com (Subhajit Sen)
wrote:
Quote:
My question is: has anybody has been able to overcome
this problem? I am aware of EKV model but do not
have access to a practical model derived from
characterizations that I could use.
Then you're out of luck. Your choices are either to bias the signal
path away from vds=0, or change to a good model.
As you say, this is a known problem. The best solution is to choose a
model which has no vds=0 kink. This implies a model which isn't
source-referenced. Both BSIM3 and BSIM4 fail. EKV mostly is OK, but
the most successful models that meet these criteria are
surface-potential based models.
--Steve
