Stats

  • Locked Locked
  • Replies 10
  • Subscribers 126
  • Views 2603
  • Members are here 0
This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

Verilog - AMS model for powerFET

Divy

To replace the spice level model with verilog ams at top level to increase the speed  .

idea here is to model powerfet  in verilog ams which should meet atleast following specs:

VDS (V) 40
RDS(on)  at VGS = 10 V 0.0063
RDS(on)  at VGS = 4.5 V 0.0075
ID (A) 58

//Verilog-AMS HDL for "FetModels_lib", "powerFET_sqj858c" "verilogams_1"

`include "constants.vams"
`include "disciplines.vams"

module powerFET_sqj858c ( D, G, S );

//==============================================================================
// Declarations (ports, parameters, variable, wires)
//==============================================================================
inout S, G, D;
electrical S, G, D;


parameter real vth = 2.0 from [0.1:10.0];
// parameter real cgs = 2.0e-9 from [1e-15:1e-6];




// variables/wires/registers
real rds;
real id, id_linear, id_sat ,rds_linear, rds_sat ;


// ===========================================================================

//==============================================================================
// Analog section
//==============================================================================

analog
begin

// when vds << vgs - vth it acts as voltage controlled resistor
//rds_sat = 50/(id_sat) ; // in sat mode the output resistance ro = 1/Id*(lamda) where lamba equals 0.02V^(-1)

//rds = V(D,S)<(V(G,S)- vth) ? rds_linear : rds_sat ;

//I(D, S) <+ V(D, S)/rds;

//I(D, S) <+ V(D, S)/rds ;

@(above(V(G,S)- vth - V(D,S),100p,50m))

begin

id = 2*2*5*( (V(G,S) - vth)*V(D,S) - 0.5*pow(V(D,S),2) ) ;

end

@(above( V(D,S) - V(G,S) + vth ,100p,50m))

begin

id = 2*2*(pow((V(G,S) - vth),2)) ; //(1+0.02*(V(D,S))) ; // drain current equation in saturation mode

end


I(D,S) <+ id ;


end
endmodule

  • Divy
    Is there any other way ,as this model is very slow
  • Andrew Beckett

    A couple of things:

    1. You don't need to use @above to do this - just use if statements for the different regions. Using @above will try to accurately solve precisely at the transition point, whereas if the curve is continuous between the regions, this should be unnecessary (and will slow the simulation down). Of course, it goes without saying that your equations should be continuous across the regions (I didn't check the maths as this is a quick response)
    2. Your code doesn't describe what id should be before it is above either of the two conditions - and also if it drops back below either of the conditions it will retain whatever previous value of id it had. I doubt that's what  you want...

    Regards,

    Andrew.

  • Divy
    I think if statements will also slow down the simulations as it will check at every time step of the simualtor
  • Divy
    Just for information i am doing parametric analysis over this model (varying both vgs and vds). I tried with this method.analog

    begin


    if (V(G,S) < vth)
    begin

    rds = 1G ;
    end

    else if (V(G,S) > vth && (V(D,S) < V(G,S) - vth ) )

    begin

    rds = 0.06/(V(G,S) - vth - V(D,S)*0.5 ) ;
    end

    else if (V(G,S) > vth && (V(D,S) > V(G,S) - vth ) )

    begin
    //id = (pow((V(G,S) - vth),2)) ;
    rds = 50/58 ;
    end

    I(D, S) <+ V(D, S)/rds ;


    end
    endmodule
  • Andrew Beckett

    Divy said:
    I think if statements will also slow down the simulations as it will check at every time step of the simualtor

    No. It's going to evaluate the code at every timestep anyway. The @above would have forced the tilmestep of the simulator, whereas if would not.

    Andrew

  • Divy

    thanks andrew for the input . I will update on this after trying what you suggested

    Regards,
    Divy

  • Divy

    So if(V(G,S) > vth ) will be faster in comparison to @above(V(G,S) - vth ) .Correct me if my understanding is wrong

  • Divy

    Hi Andrew,

    I tried to use if instead of above as suggested ,there is a difference i agree  but not much.

    In my model i observe there are sharp edges or discontinuity in curve.which can be a one reason for slowness .

    Any idea how to smooth the curve??? (any function which you can recommend)

    I tried tanh but its not solving my problem .still i see sharp edges 

    I am doing parametric anlaysis (just FYI)

    Regards

    Divy

  • Andrew Beckett

    Divy,

    If there are discontinuities in your model, then the right thing to do is to correct the equations so they aren't discontinuous - clearly the device itself doesn't exhibit discontinuous behaviour, so your regions should really join. I can't suggest smoothing options that will magically fix incorrect modelling! Using things like tanh smoothing are a good idea to ensure that the derivative is smooth between the regions, but you really need to get your basic model correct first!

    Kind Regards,

    Andrew.

Community Guidelines

The Cadence Design Communities support Cadence users and technologists interacting to exchange ideas, news, technical information, and best practices to solve problems and get the most from Cadence technology. The community is open to everyone, and to provide the most value, we require participants to follow our Community Guidelines that facilitate a quality exchange of ideas and information. By accessing, contributing, using or downloading any materials from the site, you agree to be bound by the full Community Guidelines.