include a model files for nmos and pmos into virtuoso

1. Put the contents of that document in a file, for example "massobrio.sp"
2. Create a schematic using transistors from analogLib - I used nmos4 and pmos4, but you could use the 3-terminal devices nmos and pmos if you prefer and set the bulk connections via inherited connections. The default is that nmos has its bulk as gnd! and the pmos as vdd! (If it was me, I'd use the four terminal device to start off with).
3. Set the model name on the pmos4 to be "pmos" and on the nmos4 to be "nmos" (to match the model file which uses these names - whatever word is after the .model in the file). I set w=10u and l=3u (the models have a delta-l (ld) of 0.8u which is pretty large, so the models are for a quite large geometry process). The delta-l is subtracted from both edges, so a 3um drawn length has an effective length of 1.4u. Even 3um is probably a bit aggressive for a length with such models, I think!
4. Then start ADE from this schematic and use Setup->Model Libraries. Reference the massobrio.sp file you created in step 1.
5. Add the analyses (e.g. a tran analysis) and run the simulation

I just tried this in a couple of minutes (it took me longer to write this reply) and it worked fine.

Regards,

Andrew.

Thanks for the reply. Does the model file name have to be "massobrio.sp" or "massobrio.m" , I mean I had the impression that model files should end with an extension .m?

regardss

In the old spectreS and cdsSpice interfaces you needed to have model files (in cdsSpice format) with the ".m" suffix. With spectre, the suffix doesn't matter, except that if the suffix is ".scs" it is assumed to be in spectre language, whereas any other suffix is assumed to be in SPICE language. Your models are in SPICE language, so pick anything other than ".scs". There is the "simulator lang=spice" or "simulator lang=spectre" directive that can be used to override this if needed, but you don't need that here.

Regards,

Andrew.

 Thanks for the confirmation, I tried to do the way as you suggested and the simulation happened but I got the below Notice in the output log file

Notice from spectre during initial setup

Pmos: Value of ‘kp’ and ‘u0’ are not consistent

Nmos: Value of ‘kp’ and ‘u0’ are not consistent

What is the effect of above on output and how to remove this?

Another issues is that as I observed the number of parameters in any other model file like ami06.m is 94 but the model file I am using for equal threshold voltage (attached) has only 27 parameters. So where will the issues arise in using such model file with less no. of parameters?

regards

This is a mos1 (level 1, i.e. Shichman and Hodges) model, and from the documentation:

The consistency between kp, uo, and C_ox(kp = uoC_ox) is checked according to the following rules:

• If tox is not given, C_ox is calculated from the default value of tox.

• If kp is given, uo is calculated from uo = kp/C_ox, whether uo is given or not. If this uo value is different from that specified in the model, a warning message tells you that the specified kp and uo values are inconsistent.

• If you do not give kp and uo is specified, kp is calculated from kp = uo * C_ox.

• If neither kp nor uo is given, their default values are used.

So, given that Cox=εox/Tox and εox=ε0*εr. ε0 = 8.854187817x10−12 [F/m] and εr=3.9, so since tox in the model is 0.1u, Cox=3.4531e-4, and so uo should be 3.0e-5/3.4531e-4. That's 0.08688, but since uo is entered in cm^2/V*s, and this is in m^2/V*s, we need to multiply by 100*100 - so that makes it 868.8 - which is exactly what you see in the model parameters output from spectre (if you've saved them - in ADE you can see this in the results browser). The model has 700 for uo. All it means is that 700 is ignored and 868.8 used instead.

The number of parameters being different may just be a matter of this being a level 1 model - I have no idea what the "ami06" model is. Also, this is a made up model representing an ancient technology - presumably the ami06 model is a characterized model from an AMI process, so has probably had more care put into its creation.

Regards,

Andrew.