Value of lambda in cadence virtuoso

[MahmoudHassan]

Hello
I am trying to know the vale of lambda to calculate the output impedance and gm required for specific gain
i tried to find it through mosfet model parameters as I know it is PCLM parameter
but I found it very very large and I think it is not logical as
Lambda for nmos = 1.815 and lambda for pmos is = 3 !!!
are these numbers in mili V^(-1) or what ?

best regards,
Mahmoud

 

[dick_freebird]

You're talking lambda in the model statement, as channel
length modulation, right? Not lambda as a PDK variable
for process-independent transistor sizing?

You could probably find the answer according to Cadence
in the Spectre modeling manual (or the deeper half of the
user manual perhaps, but not always the detail you seek).
Dig deep enough and you might even find an equation;
if this is BSIMx.y then all of that should be public domain.

Lambda should produce a dIDsat/dVds slope that matches
real device drain I-V curves. It will almost never dead
match across geometries and process corners and so on,
unless it's made to be an expression (you show constants
and so this is likely a digital kit). Model "LEVEL" to "LEVEL"
I recall seeing the units basis change, as years went by,
and I wouldn't care to speculate what the units basis is
in your case.

 

[erikl]

Lambda for nmos = 1.815 and lambda for pmos is = 3 ! are these numbers in mili V^(-1) ?

PCLM is a dimensionless channel length modulation parameter, it belongs to the denominator in the CLM-dependent part of the Early Voltage (V_ACLM) equation, see e.g. any MOSFET models or BSIM manual.

 

[MahmoudHassan]

Thanks a lot for your interest and care and Happy New Year
Actually it is being more difficult!
Ok, But if I want now to design differential amplifier the simple one with gain of 100 and current of say 10uA
then I need to design for specific gm and Rout (Rout= r01 || r02)
How can I know the output resistance (I need at least lambda ) for specific tech file say tsmc130nm?
and If need to use recent model like BSIM4 How can I start my design with these new model equations?
Best Regards

 

[erikl]

How can I know the output resistance (I need at least lambda ) for specific tech file say tsmc130nm?

Channel length parameter lambda is only used in some low level (level=1, 2, 6, 7, 8) spice model files (and with different units: 1/V , cm/√V , or without a unit), in the EPFL-EKV model level=55 (without a unit), as well as in some of the newer - very dedicated - BSIM4 levels (61, 62, and 72), see e.g.

If need to use recent model like BSIM4 How can I start my design with these new model equations?

Get foundry model files e.g. from MOSIS - against registration via a University / High School license account, or use the PTM model files (scroll down).

Happy New Year to you too!

 

[MahmoudHassan]

Channel length parameter lambda is only used in some low level (level=1, 2, 6, 7, 8) spice model files (and with different units: 1/V , cm/√V , or without a unit), in the EPFL-EKV model level=55 (without a unit), as well as in some of the newer - very dedicated - BSIM4 levels (61, 62, and 72), see e.g.


Get foundry model files e.g. from MOSIS - against registration via a University / High School license account, or use the PTM model files (scroll down).

Happy New Year to you too!

Greate, If I want to start designing differential pair and for examble I want to get a gain of 100
How can precisily design bias circuit (VGS, ID ) and gm and ro to get this gain with these advanced models
I tried using conventional ones but the results were waay from the hand calculated

 

[erikl]

How can precisily design bias circuit (VGS, ID ) and gm and ro to get this gain with these advanced models
I tried using conventional ones but the results were waay from the hand calculated

This is always so, if you work with sub-micron processes: too many influences aren't considered by the hand calculations. I'd recommend to use the gm_over_Id design methodology for sub-micron designs. Search for gm/Id in this forum, or in G00GLE, you'll find a lot of contributions. To study this method from scratch, I'd recommend this book.

 

[MahmoudHassan]

Thanks a lot Erikl
God bless your efforts
I started to learn this methodology
It is really hard

 

[erikl]

I started to learn this methodology It is really hard

Only in the beginning. Just don't give up! Good luck!