The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design Languages >> Verilog-AMS >> Help with Verilog -A：bandpass lna
https://designers-guide.org/forum/YaBB.pl?num=1196252195

Message started by surer on Nov 28^th, 2007, 4:16am

Title: Help with Verilog -A：bandpass lna
Post by surer on Nov 28^th, 2007, 4:16am

I want to realize a model of fo - >1GHz and bandwidth - > 100MHz low noise amplifier. But Ican't realize the bandpass characteristic and the flat bandwidth. Maybe the lowpass easy to realize?
post the code,and help me to modify.

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

module lna_single(rf_in,rf_out,gnd);
inout rf_in,rf_out,gnd;
electrical rf_in,rf_out,gnd;
electrical p1,p2,p3;

parameter real gain = 10; //power gain in dB
parameter real r_in = 50 from (0:10K]; //input resistance in Ohms
parameter real r_out= 50 from (0:1M]; //output resistance in Ohms
parameter real iip3 = 10; //input referred 3rd order intercept point in dBm
parameter real nf = 2.0 from (0:inf); //noise figure in dB
parameter real fg = 1100M from [1M:20000M];
parameter real fc = 900M from [1M:19000M];
parameter real fb = 700M from [1M:19000M];
parameter real fa = 1300M from [1M:20000M];

real a,c,rnf,gain_lin,ip3_lin;
real inmax,outmax;
real noise_voltage_squared;
real zeros[0:1],poles[0:1];

analog begin
//Convert the input parameter from engineering units to implementation units
@(initial_step) begin
gain_lin = pow(10,gain/10);
ip3_lin = sqrt((pow(10,iip3/10))*2*r_in*0.001);
rnf = pow(10,nf/10);
noise_voltage_squared = 4*`P_K*(rnf-1)*$temperature*r_in;

a = sqrt(gain_lin*r_out/r_in); // voltage gain
c = (4*a)/(3*ip3_lin*ip3_lin); //

// Compute the critical point
inmax = sqrt(a/(3*c));
outmax = (2*a*inmax)/3;

// Define transfer function zero and pole
zeros[0] = `M_TWO_PI*fa;
zeros[1] = `M_TWO_PI*fb;
poles[0] = `M_TWO_PI*fc;
poles[1] = `M_TWO_PI*fg;
end

//add noise voltage source
V(rf_in,p1) <+ white_noise (noise_voltage_squared,"lna_diff");

//input resistance
V(p1,gnd) <+ I(p1,gnd)*r_in;

//frequency response & consider the function laplace_zp
V(p2,gnd) <+ laplace_zp(V(p1,gnd),zeros,poles);

//nonlinear characteristic
if(abs(V(p2,gnd))<inmax)
V(p3,gnd) <+ 2*(a-c*V(p2,gnd)*V(p2,gnd))*V(p2,gnd);
else if (V(p2,gnd) > 0)
V(p3,gnd) <+ 2*outmax;
else
V(p3,gnd) <+ -2*outmax;

//output resistance
V(rf_out,p3) <+ I(rf_out,p3)*r_out;

end
endmodule

Title: Re: Help with Verilog -A：bandpass lna
Post by Geoffrey_Coram on Nov 28^th, 2007, 7:53am

I know some simulators require the zeros and poles arguments to be constants; you have yours set up as variables that are set in the @initial_step. (They're variables, even though you set them from constants/parameters.)

Try using a fixed "vector" {1, 2, 3} for zeros and poles and see if this works, then we can investigate ways to use parameter arrays.

Title: Re: Help with Verilog -A：bandpass lna
Post by surer on Nov 28^th, 2007, 4:41pm

my lna verilog-a code is refered to the book "Modeling.and.Simulation.for.RF.System.Design".
the main question locates in the laplace function.
and I dont know how to realize the 100M flatten gain. If use the zeros and poles, need enough Octave to satisfy the code function.
in addition, the bandpass is other issue to me. All paper provides the method to realize lowpass. I try to use the method of ative filters to realize the bandpass and not succeed.
So maybe the code needs to use new techique or modify the method that I used. help me and I starve for the behavioral models to AMS simulation. 3x