MOS - Width and Lenght dimensions in Spice Simulation

[tonyoxy]

Hi everyone,

I'm trying to simulate a simple electric circuit using spice with a common MOS insite (i.e. ZVN4424Z - SOT89 N-CHANNEL ENHANCEMENT MODE VERTICAL DMOS FET).
The supplier Zetex (https://www.diodes.com/) give as the MOS spice model.
The question is:

I'm quite sure (see in the follow) that the model doesn't consider the actual size of the MOS (i.e. the width W and the lenght L).

Is it possible to derive these parameters from the datasheet? any suggestions?

Here I report the Spice Model suggested from ZETEX. In my opinion it is evidente that the dimension W and L are not considered into the model.

Marco


*
*Zetex ZVN4424Z Spice Model v1 Last Revised 28/7/08
*
.SUBCKT ZVN4424Z 3 4 5
*------connections-------D-G-S
M1 6 2 8 8 N4424AM
M2 6 2 8 8 N4424AMS
RG 4 2 35
RD 3 6 3.6
RS 8 5 0.15
RL 3 5 240E6
D1 5 3 N4424AD
Egs2 13 8 2 8 1
Eds1 14 8 6 8 1
C1 2 8 145E-12
C2 2 3 11E-12
C3 15 14 235E-12
C4 16 8 226E-12
S1 2 15 14 13 SMOD1
S2 13 15 14 13 SMOD2
S3 16 13 13 8 SMOD3
S4 16 2 13 8 SMOD4
.MODEL SMOD1 VSWITCH RON=.001 ROFF=100 VON=1 VOFF=2
.MODEL SMOD2 VSWITCH RON=.001 ROFF=100 VON=2 VOFF=1
.MODEL SMOD3 VSWITCH RON=.001 ROFF=100 VON=-3 VOFF=-4
.MODEL SMOD4 VSWITCH RON=.001 ROFF=100 VON=-4 VOFF=-3
.MODEL N4424AM NMOS VTO=1.3 IS=1E-15 KP=0.85
+CBD=66.2E-12 PB=1
.MODEL N4424AMS NMOS VTO=.86 IS=1E-15 KP=0.0085
+CBD=.66E-12 PB=1
.MODEL N4424AD D IS=5.516E-13 RS=.2084 N=1.0078
.ENDS ZVN4424Z

 

[zizu]

Since this is a discrete component, I suppose the dimension is not important here.

 

[tonyoxy]

Hi zizu,
thanks for reply.

Why you made a differenze fron discrete anda non discrete components?
The MOS part is always inside the package and the mathematical MOS equations are still the same.

Is it right?

 

[zizu]

Hi tony,
Yes, you are absolutely right.
However, you don't have to know the dimension if you are to "use" it since you can't change the dimension. The model is built for people to use it and thus dimension is not considered in it.
But if you are trying to realize this component from inside, it might be important for you to know the dimension. BTW, I guess you need to know at least the process in which this chip is fabricated to derive its dimension.

 

[tonyoxy]

Hi zizu,

I have to design a circuit starting fron simulation.
Due to the fact that I see some differences from simulation and measure I want to push some inside in the model I have used.
You know that mathematica equations of MOS depends on W/L ratio. PSpice implements that equations and, by default PSpice use W=100um and L=100um for MOS dimention parameters.

So, my question is how can I use a more accurate values of W and L from datasheet or fron package imformation (SOT 89 in my case) that, we know, depends from the fabrication's process chip.

 

[zizu]

Hi tony,
So you are trying to build a new accurate model based on the original inaccurate one? or to apply the real W/L ratio to this model? The model suppose to be fitted to its measurement results even if its W/L in it is not correct. If, unfortunately, it's not well built, how could you trust this model and build or modify it from itself. That's why I don't think knowing the dimension could help you in this case.

 

[tonyoxy]

Zizu,

I don't want to build a new model.
I want only apply the real W/L ratio to my simulation in the ZETEX model and I don't know how can i do that.

In my previous post I said that "some differences" from measured and simulation appear but I don't see a completely different results so I think that the model is well built.

 

[zizu]

Just for my curiousity, if any evidences there convince you that W/L cause the differences between your simulation and measurement? You may also have to check if this model (it seems a low level model and has already taken w,l into one of its parameters) takes W,L as its arguments. If not, you can't apply W,L to it. If you really care the difference, you probably want to build a new model for it by conducting some measurement by yourself.

 

[tonyoxy]

Zizu,

I'm sure that W/L ratio influences the simulation results because when I tried to do this on model (passing W and L as parameters) the results changed.
In any case I don't want follow the "try and error criteria" i.e.:
1) change W and L,
2) obtain the simulation results,
3) compare these results with measurements
4) change again W and L,
.
.
n).... and so on.

I would be sure of W and L values taht I insert in the model.
And I'm looking for a way of finding these starting from datasheet or package or process fabrication information.

 

[edge_tv]

I looked into the Datasheet. I thinke you have to fit your model by hand, even if you know the correct W and L.
The Vth can go from 0.8 V- 1.8V ! The model uses the nominal value 1.3V. I would try to measure Vth from the real MOS to see if Vth is your problem.

 

[tonyoxy]

Hi edge_tv,

I think what you said was right.
But in my opinion the VTh voltage tolerance can be taken into account using, for example, Montecarlo Simulation method in spice using the information you have found inside datasheet.
The problem still remains the physical dimension W and L because I have no ideas about the values I must use.