FET Noise in BSIM4

Dear illaoi,

illaoi said:

I am trying to find gamma in a single FET simulation to match current noise density (maybe op from calculator), in^2=4*k*T*gamma*gm. I did spectre -h bsim4 but not sure where to find it

I have not seen the gamma factor as a noise parameter in the BSIM4 model. It can vary significantly depending on the channel length (as high as 7 from reference [1] shown below) and is difficult to measure experimentally. Long channels devices do possess an excess noise factor close to the expected value of 2/3. However, to my knowledge, it is not a BSIM4 parameter.

Shawn

Figure 1

(from reference [1])

[1] R. Navid, R. Navid, T. H. Lee and R. W. Dutton, "A Circuit-Based Noise Parameter Extraction Technique for MOSFETs," 2007 IEEE International Symposium on Circuits and Systems, 2007, pp. 3347-3350,

Thanks Shawn, also I was wondering if you can comment on following regarding noise figure. This is what I found with spectre -h noise

I just don't understand the term "Nl" here, any comment?

The noise analysis (it's similar with pnoise and hbnoise) compute the amount of noise at each noise source (which is bias dependent in some cases) and the transfer function from each noise source to the output. Consequently it computes the contribution from each noise source at the output. The total output noise is the sum of the noise powers from each contributor at the output. 

Nl is the noise contribution caused by the output probe (i.e. the load). Ns is the noise contribution caused by the input probe (i.e. the port at the input). So what the noise factor (F) is the total noise at the output, less the noise from the output load (which can be considered the probe that is part of the "test equipment", divided by the noise contribution from the input source. In effect this is the ratio of the signal to noise at the input to the signal to noise at the output - which tells you the increase in noise in the circuit (the signal effectively cancels out by computing it this way).

Andrew

Dear illaoi,

illaoi said:

I just don't understand the term "Nl" here, any comment?

Let me hope to understand your question and, hopefully, provide some insight. The Nl  term is provided after a noise analysis is performed and represents the portion of the noise attributable to the output load. From the expression for NF and F, it is evident that Nl is removed from the total noise - which makes sense as the NF for a amplifier should not include the impact of the noise associated with the load.

Andrew Beckett has assembled a few sample test cases that you might examine that include a noisy output port in the Article at URL:

https://support.cadence.com/apex/ArticleAttachmentPortal?id=a1Od0000000nSkyEAE&pageName=ArticleContent

I hope this explanation and Andrew's examples help to clarify the definition of NI illaoi.

Shawn