The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> Thermal Noise in Spectre
https://designers-guide.org/forum/YaBB.pl?num=1066831951

Message started by mateenfl on Oct 22^nd, 2003, 7:12am

Title: Thermal Noise in Spectre
Post by mateenfl on Oct 22^nd, 2003, 7:12am

Hello,

The gate referred thermal noise density for MOS is given in many books as 4KT(2/3*gm).

It is also mentioned this equation is valid only for long channel devices. For short channel devices, this is mentioned to be 2 to 5 times larger (called the excess noise factor)

Does Spectre take into account this excess noise factor for short channel devices e.g. at 0.18u, when performing noise analysis? (How can I determine this excess noise factor through device models/simulations etc?)

Kind Regards,
Faisal Mateen

Title: Re: Thermal Noise in Spectre
Post by Ken Kundert on Oct 22^nd, 2003, 10:22am

The factor of 2/3 is a model parameter that can be set in most of Spectre's MOS models. Currently I am away from my desk and cannot tell you the name of the parameter, but if you run "spectre -h model", where model is the type of the MOS model you are using (bsim3v3, etc.), and look at the noise parameters, you should be able to find it easily.

-Ken

Title: Re: Thermal Noise in Spectre
Post by skippy on Oct 22^nd, 2003, 12:21pm

The 2/3 factor in the expression is analytically derived for the onset of saturation. The factor approaches 1 as VDS approaches 0. I don't believe analytical expressions exist beyond saturation, but please let me know if you are aware of any.

At short channel lengths there are many additional effects. As gate oxides get thinner, there is also shot noise through the gate.

A derivation by van der Ziel appears in his 1986 book, "Noise in Solid State Devices and Circuits." There are several papers which discuss the equivalence of his model with those used in BSIM.

The excess noise factor could be best found through measurement or numerical device simulation. I would suspect that this factor varies with both gate and drain bias.

Title: Re: Thermal Noise in Spectre
Post by Paul on Oct 23^rd, 2003, 12:13am

The additional noise in short-channel devices is due to velocity saturation and hot-carriers. As far as I know, the effect of hot carriers on the thermal noise is still discussed today and some people disagree. Some papers on the subject:

P. Klein, "An analytical thermal noise model of deep submicron MOSFETs", IEEE Electron Device Lett., vol. 20, pp. 399 401, Aug. 1999.
http://www.cr.org/MSM2002/program/WCM2002tut-Enz.pdf
http://www.unibw-muenchen.de/campus/ET4/publikationen/knob101.pdf
http://www.ece.mcmaster.ca/~chihhung/Publication/ED2002_ChannelNoise.pdf

Some people also consider gate induced noise, only modeled in BSIM4 AFAIK.

Paul

Title: Re: Thermal Noise in Spectre
Post by qiushidaren on Jul 16^th, 2007, 9:35am

By the way Paul, what's AFAIK stand for here?

-Terry

Title: Re: Thermal Noise in Spectre
Post by Ken Kundert on Jul 16^th, 2007, 2:13pm

AFAIK = as far as I know

Title: Re: Thermal Noise in Spectre
Post by qiushidaren on Jul 16^th, 2007, 6:47pm

Thanks Ken, haha, I thought it standed for some modeling techniques or something, I am not a native speaker.

-Terry