about parameter lamda

[lhlbluesky]

in 0.35um or 0.18um process, how to extract the parameter of lamda?
in simulation,i find that lamda changes with various factors,such as Vgs\vds\W and so on,and the value varies very sharply,but how to calculate rout using lamda for different operating conditions?
besides,in spectre,there is a parameter ron,what's the difference of ron and rout?
how to calculate dc-gain using ron?

 

[sunjunjun]

just use a simple model
for example,use DC analysis
lamda=d(i)/d(v)

 

[lhlbluesky]

lamda=d(i)/d(v),what's the value of v?
v=vds or vgs?

 

[yanivh]

The transistor output resistance is ro=1/(lambda*Id) hence by measuring Id and ro you got lambda.
In addition by defenition lambda is propotonal to dXd/dVds.

 

[amitjagtap]

I suggest u to use different methods to find out lambda. and find out average of the most common values. its better to use values arround ur required Vds.
the diif methods are
1) slope= lambda/Id' where Id' is the current at which tangent to the curve intercept to the y axis.
2) select two Vds values,mostly in deep saturation because in deep saturation lambda is more accurate.
then use
Id2/Id1= (W/L)2/(W/L)1 * (1+Lambda.Vds2)/(1+Lambda.Vds1)
3) Plot a graph of Rout Vs Vds this curve has shape like gaussian curve.
From which u will get VA@DIBL and VA@CLM
Then VA(Early voltage)= VA,Sat + 1/(1/Va,DIBL + 1/Va.CLM)
for more details of this refer some MOSFET model book. also given in BSIM manual.
Hope this will help u.:idea:

 

[lhlbluesky]

thanks all.

 

[khaled ashraf]

hii everyone ,
I have a question when I tried the method of lamda = d(I)/d(Vds) at L=60n
I found that for nmos lamda =0.3345 while when I tried the other method of
I1/I2=[(W/L)1*(1+lamdaVds)]/[(W/L)2*(1+lamdaVds)] I found that it became arounf 0.65 and when I tried ro=1/(lamda*Id) I got variations from 0.12 to 0.3
and the value depends on Vgs,Vds,W
so what shall I do???:-(