The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design Languages >> Verilog-AMS >> Antenna with hidden state
https://designers-guide.org/forum/YaBB.pl?num=1151412935

Message started by Peterraus on Jun 27^th, 2006, 5:55am

Title: Antenna with hidden state
Post by Peterraus on Jun 27^th, 2006, 5:55am

When building an antenna in VerilogA, the simulation work for tran, ac and sp, but for pss, spectre indicates a hidden state of the variables L and C.

The antenna has an available power (receiving a signal at a certain frequency) and an impedance. The impedance is determined by the magnitude and angle of the reflection coefficient. In this way, a constant magnitude and swept angle represents all impedances that a circle with constant radius passes through in the smith chart. (A circle in the Smith chart corresponds to a constant VSWR value).

With this construction of the antenna, depending on the value of the angle, the impedance should be represented by a capacitor (lower part of Smith cart) or an inductor (upper part of Smith chart) in series with a resistor and a voltage source. I have put the inductor and capacitor in parallel and the angle determines which is conducting.

I have tried to "activate" the L and C for every timestep but, perhaps, since the angle is independent of the time it has not been a success.

Any ideas how to solve or completely eliminate the hidden state problem?

// VerilogA for DVBUnitcells, Antenna, veriloga

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

module Antenna(v_ref, feed);

inout v_ref, feed;

///////////////////////
// v_ref is ground.
// between ground and v is a voltage source v_source
// between v and res is a resistor with value R
// Between the resistor and the antenna feed is a capacitor or a inductor.
//////////////////////

electrical v_ref, v, res, feed;

parameter real AvailablePower_dBm = -50 from (-inf:inf);
parameter real Magnitude = 1m from [0:1);
parameter real Angle_rad= 1m from [0:6.2832];
parameter real Frequency = 1G from (0:inf);

//local variables
real AP, mag, ang, f, v_source, R, L, C, pi ;

analog begin

pi=3.14159265358979323846264338327950288419716939937511;

AP = AvailablePower_dBm;
mag = Magnitude;
ang = Angle_rad;
f = Frequency;

if (mag == 0) begin
mag = 1n;
end

if (ang == 0) begin
ang = 1n;
end

if (ang>6.2831) begin
ang = 6.2831;
end

if (ang>3.1415 && ang<3.1416) begin
ang = 3.1415;
end

// x^y is in Skill x**y but in VerilogA pow(x,y).

v_source = sqrt(1E-3*pow(10,(AP/10))
*8*(-50*(-1+pow((mag*sin(ang)),2)+pow((mag*cos(ang)),2))/
(1+pow((mag*sin(ang)),2)-2*mag*cos(ang)+pow((mag*cos(ang)),2))));

R = -50*(-1+pow((mag*sin(ang)),2)+pow((mag*cos(ang)),2))/
(1+pow((mag*sin(ang)),2)-2*mag*cos(ang)+pow((mag*cos(ang)),2));

// Inductor and capacitor is parallel
// but given values in a way that only one is conducting current

if (ang<=3.1415) begin // Inductor
L =100*mag*sin(ang)/((1+pow((mag*sin(ang)),2)-2*mag*cos(ang)+
pow((mag*cos(ang)),2))*2*pi*f);
C = 1a; //small but nonzero
end

if (ang>=3.1416) begin //Capacitor
C = 1/(2*pi*f*(-100*mag*sin(ang)/
(1+pow((mag*sin(ang)),2)-2*mag*cos(ang)+pow((mag*cos(ang)),2))));
L = 1; //Large
end

V(v,v_ref) <+ v_source*sin(2*pi*Frequency*$abstime);
V(v,v_ref) <+ ac_stim("ac",v_source); // ac only
V(v,v_ref) <+ ac_stim("pss",v_source); // pss only
V(v,v_ref) <+ ac_stim("sp",v_source); // sp only
I(v,res) <+ V(v,res)/R;
I(res,feed) <+ idt(V(res,feed)/L);
I(res,feed) <+ ddt(V(res,feed)*C);

end //analog end
endmodule

Thanks

Peter Sjöblom
peter.sjoblom@es.lth.se

Title: Re: Antenna with hidden state
Post by Geoffrey_Coram on Jun 27^th, 2006, 6:32am

Yikes! what a mess.

I would recommend that you change the two if statements
if (ang<=3.1415) begin // Inductor
if (ang>=3.1416) begin //Capacitor

into one if/else statement. Maybe then you can also remove that bizarre
if (ang>3.1415 && ang<3.1416) begin
ang = 3.1415;
end

You should also look into `M_TWO_PI from constants.vams instead of "6.2832" (in the range for Angle_rad) and instead of 2*pi in some expressions.

Why do you set ang=1n if it is == 0??

I would also recommend creating variables to hold some sub-expressions such as cos(ang) and sin(ang); also, pow() is quite expensive, so you'd be better with
mag_x = mag * cos(ang);
mag_y = mag * sin(ang);
magx2 = mag_x * mag_x;
magy2 = mag_y * mag_y;

Depending on the simulator you have, you may be able to create localparams,
localparam sinang = sin(Angle_rad);
Localparams can't be overridden in the netlist, and their values are set once at the start of the simulation, so you don't spend time re-computing their values on every iteration of every timepoint.

I don't think ac_stim("pss", v_source) will have any effect, since pss is not a small-signal analysis. Perhaps you want ac_stim("pac", v_source)?

Wait, are you sure you want v_source and not 1.0? Eg, when you do a regular voltage source, you don't generally want the dc voltage to be the same as the ac magnitude.

I would also stronly recommend changing the inductor formulation to use ddt() instead of idt(); you'll need a switch branch, but it's still more natural for a Spice-like simulator. Otherwise, you probably need to provide an initial condition for the idt() operator.