[SOLVED] early voltage of tsmc018 process, please help me

[arafatsagar]

early voltage

i have done extrapolating the Id versus Vds curves (shown in the attached file) to get the early voltage. But unfortunately i have got no early voltage. the data of Id,Vds,Vgs have been extracted from HSPICE simulation. Then i have done the extrapolation by MATLAB. But the curves do not intersect at any single point to produce the early voltage. please help me. if it is not possible to get the early voltage by extrapolation, please help me telling the way of getting the early voltage.



if i predict second order dependence of Id on Vds rather than a linear relationship in saturation the extrapolation becomes



but it is not satisfactory. please help me. thanking in advance.

 

[LvW]

early voltage site:www.edaboard.com

As FvM has mentioned already
"Early voltage is a BJT parameter" - and has no practical meaning for MOSFETs.
(More than that, also for BJTs it is only a rough approximation with the aim to model the real transistor behaviour)

Why do you want to find something like an Early voltage for MOSFETs ?
It makes really no sense as such a parameter does not exist !

 

[arafatsagar]

early voltage

thank you very much for your reply.
its a portion of my thesis work and it is urgent.
why do MOSFET have no practical meaning for early voltage?
is Va = 1/λ only an approximation for ideal case? has it no meaning at all?
in fact according to my two uploaded images i have not got any particular intersecting point for early voltage. are those images reflecting the real situation?
it will be a great help for me to get your valuable feedback.

 

[LvW]

what is early voltage

...............................
in fact according to my two uploaded images i have not got any particular intersecting point for early voltage. ....

Why do you expect such an intersecting point ?
(remember: in reality, it also does not exist for BJTs !)

 

[arafatsagar]

early voltage mosfet definition

thanks for the information. will you please prescribe me any book or paper where i can get more information about the non-existance of early voltage for MOSFETs? i need to prove that there is nothing like early voltage for MOSFETs. is the geometrical proof is enough?

 

[LvW]

tsmc018

thanks for the information. will you please prescribe me any book or paper where i can get more information about the non-existance of early voltage for MOSFETs? i need to prove that there is nothing like early voltage for MOSFETs. is the geometrical proof is enough?

In general, it is not easy to prove that something does not exist.
Do you have any measurements available ?
What about simualtion results on the basis of detailed simulation models ?
(You know, for MOSFETs there a several models available)

 

[arafatsagar]

early voltage dependency

actually, i used simulation result of BSIM3 VERSION 3.1 Level = 49 to get the Id versus Vds curve in saturation. then i extrapolated to get Va, but i had got nothing.
what is an early voltage actually?
If a solid state element has an early voltage, what physical significance does the device has?
Why doesn't a MOSFET have that physical significance?
thanks a lot in advance.

 

[LvW]

mosfet early voltage

....................
If a solid state element has an early voltage, what physical significance does the device has?
Why doesn't a MOSFET have that physical significance?
thanks a lot in advance.

In fact, the Early voltage has NO physical significance. That means, it cannot be measured directly (as a voltage) nor can it be derived from mechanical properties.
The only thing Mr. Early has discovered is, that the extrapolated curves Ic=f(Vce) more or less meet in a common point on the negative Vce axis. And this simplified property can be used for calculation purposes (when the slope of the curves is needed). That´s all. It is purely a fictive voltage of tens of volts.

With some additional assumptions and simplifications one can transfer this theoretical picture from BJTs to MOSFETs and define something similar also for MOSFETs, But, as mentioned earlier, the error in this case is larger - and I don´t think that the different levels for MOSFET models use such a voltage to model the Id=f(Vds) characteristics.
When your curves do not meet in a common point - trust your measurements/simulations and argue that definition of such a voltage makes no sense since it does simply not exist !

 

[FvM]

early voltage mosfet

To add a simple consideration: The BJT early voltage isn't but a quantity, that represents the roughly reciprocal dependency between Ic and differential output resistance in common emitter circuit. ro = VA/Ic. Although it doesn't represent the exact behaviour, it's a good first order estimation. MOSFET also have a current dependant differential output resistance, but it's obviously very different for real devices.

P.S.: Ic in the above equation is the extrapolated value at Uce=0.

 

[arafatsagar]

early voltage calculation for mosfet

thanks a lot LvW and FvM.
i have got my answer from the discussion. actually, i tried to get the early voltage for real BJTs but they always have error. in case of MOSFETs the %of error is quite appreciable. if my measurements and simulations are correct, i think early voltage is just a conception used for modeling the classical ideal BJTs. thanks a lot for helping me.