The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design Languages >> Verilog-AMS >> parameterize the dimension of variables
https://designers-guide.org/forum/YaBB.pl?num=1198040383

Message started by Aigneryu on Dec 18^th, 2007, 8:59pm

Title: parameterize the dimension of variables
Post by Aigneryu on Dec 18^th, 2007, 8:59pm

Hi,

In verilog, we can parameterize thedimension of the bus, but I found it may not be allowed in verilog-a. Is this true?

In verilog

module abc(out, in);
parameter bitnum=3;
output [bitnum-1:0] out;
input [bitnum-1:0] in;

endmodule

But I cannot do this in verilog-a. Is there any way to do this?
I need to create an array of electrical signals while I wish to parameterize the dimension.
Is there a way to achieve the following function?

parameter integer dimension=3;
electrical [dimension-1:0] node_voltage;

Title: Re: parameterize the dimension of variables
Post by Ken Kundert on Dec 19^th, 2007, 10:15am

This is not a limitation of Verilog-A, rather it is a limitation of the simulator. Are you using Spectre? Spectre has suffered from this problem since Verilog-A was introduced. I can sometimes work around the problem by using 'defines.

-Ken

Title: Re: parameterize the dimension of variables
Post by Aigneryu on Dec 20^th, 2007, 2:55pm

Thanks Ken,

That's right, I am using spectre in cadence. If I change to verilog-AMS, will it help?
If I change to verilog-AMS, will it require more license compared to verilog-A (license for IUS may be required).

Sincerely,

Title: Re: parameterize the dimension of variables
Post by Aigneryu on Dec 20^th, 2007, 2:58pm

Actually my question is how to make the following code more flexible in terms of parameter. This is a Gaussian filter which I want to use in analog simulation.

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

module FIR4 (out, in, gnd, phi1, phi2);
output out;
electrical out;

input in;
electrical in;
input gnd;
electrical gnd;

input phi1;
electrical phi1;
input phi2;
electrical phi2;

parameter real vth1 = 0.0 ;
parameter real vth2 = 0.0 ;
parameter real ctap = 10p ;
parameter real roff = 1G ; //ohm
parameter real ron = 0.1 ; //ohm
parameter real osr = 16;
parameter real fc_anti_alias = 4M;

electrical [0:32] vphi1, vphi2, vc1, vc2;
electrical vsum1, vsum2;
real rphi1, rphi2;
real tap_coeff [0:32];
real c_anti_alias;

genvar m, n, p;

analog begin
@(initial_step)
begin
// initial condition of switches
rphi1=roff;
rphi2=ron;
// set cap for anti alias filter
c_anti_alias=1/(2*3.14159265358979*fc_anti_alias*100);

// load tap coefficients
tap_coeff [0] = 0.126951 ;
tap_coeff [1] = 0.157485 ;
tap_coeff [2] = 0.192665 ;
tap_coeff [3] = 0.232450 ;
tap_coeff [4] = 0.276577 ;
tap_coeff [5] = 0.324536 ;
tap_coeff [6] = 0.375553 ;
tap_coeff [7] = 0.428589 ;
tap_coeff [8] = 0.482361 ;
tap_coeff [9] = 0.535382 ;
tap_coeff [10] = 0.586026 ;
tap_coeff [11] = 0.632602 ;
tap_coeff [12] = 0.673451 ;
tap_coeff [13] = 0.707037 ;
tap_coeff [14] = 0.732048 ;
tap_coeff [15] = 0.747477 ;
tap_coeff [16] = 0.752692 ;
tap_coeff [17] = 0.747477 ;
tap_coeff [18] = 0.732048 ;
tap_coeff [19] = 0.707037 ;
tap_coeff [20] = 0.673451 ;
tap_coeff [21] = 0.632602 ;
tap_coeff [22] = 0.586026 ;
tap_coeff [23] = 0.535382 ;
tap_coeff [24] = 0.482361 ;
tap_coeff [25] = 0.428589 ;
tap_coeff [26] = 0.375553 ;
tap_coeff [27] = 0.324536 ;
tap_coeff [28] = 0.276577 ;
tap_coeff [29] = 0.232450 ;
tap_coeff [30] = 0.192665 ;
tap_coeff [31] = 0.157485 ;
tap_coeff [32] = 0.126951 ;

end

rphi1 = ((V(phi1) > vth1)? ron:roff);
rphi2 = ((V(phi2) > vth2)? ron:roff);

for (m=0; m<=32; m=m+1)
begin
//sampling
I(vphi1[m], vc1[m]) <+ V(vphi1[m], vc1[m])/rphi1;
I(vphi2[m], vc2[m]) <+ V(vphi2[m], vc2[m])/rphi2;
//integration
V(vc1[m], gnd) <+ idt(I(vc1[m], gnd)/ctap, 0);
V(vc2[m], gnd) <+ idt(I(vc2[m], gnd)/ctap, 0);
//following
V(vphi2[m])<+V(vc1[m]);
//filtering
V(vsum1)<+tap_coeff[m]*V(vc2[m])/osr;
end

V(vphi1[0])<+V(in);

for (n=1; n<=32; n=n+1)
begin
V(vphi1[n])<+V(vc2[n-1]);
end

I(vsum1, vsum2) <+ V(vsum1, vsum2)/100;
V(vsum2, gnd) <+ idt(I(vsum2, gnd)/c_anti_alias, 0);

V(out, gnd) <+ V(vsum2, gnd);

end
endmodule

Title: Re: parameterize the dimension of variables
Post by Ken Kundert on Dec 20^th, 2007, 10:36pm

Seems like a very expensive approach. Why don't you just use the z-domain filter functions?

-Ken

Title: Re: parameterize the dimension of variables
Post by Aigneryu on Dec 21^st, 2007, 1:00pm

The reason I use this is because for some reason I need to play with the dual phase clocks. And I will also need to make a smooth transition for the switches in order to pass the small signal from the clock ports.