The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Modeling >> Behavioral Models >> Solving a differential equation using VerilogA Model
https://designers-guide.org/forum/YaBB.pl?num=1379599562

Message started by Ricky Chen on Sep 19^th, 2013, 7:06am

Title: Solving a differential equation using VerilogA Model
Post by Ricky Chen on Sep 19^th, 2013, 7:06am

Dear All,

I need to solve a differntial eqaition in my VerilogA marco-model

The eqaution I want to solve is:-

ddt(V(alpha)) = Vx * V(in)

V(in) is the input signal ( as a function of time 't') to the model.

Vx is read from a "table_model" at (t + V(alpha) i.e ( $(abstime) + V(alpha) )

To solve this I did as below:-

// define branches
branch (alpha) alpha1;
branch (alpha) alpha2;

I(alpha1) <+ -Vx * V(in);

I(alpha2) <+ ddt(V(alpha));

The KCL at alpha forces I(alpha1) + I(alpha2) = 0 and solve the differntial equation.

I went and checked the results.

At $abstime = t1

Vx * V(in)= x1

V(alpha)= y1

At $abstime = t2

Vx * V(in)= x2

V(alpha)= y2

I think y2 should be equal to

y2= [ x2 * (t2 - t1) ] + y1.

But SPECTRE is showing something different than "y2" .

Can anybody please tell where I am going wrong ?
Whether my line of thought is correct ?
or it may be some accuracy issue of Spectre lile Reltol, Vabstol,Abstol playing some rule ?

Also I solve the above differntial eqaution using MATLAB ODE45().
It is totally different from the SPECTRE result !

Kind Regards,

Title: Re: Solving a differential equation using VerilogA Model
Post by boe on Sep 23^rd, 2013, 12:52am

Ricky Chen wrote on Sep 19^th, 2013, 7:06am:

[Added:] dValpha(t)/dt = f(Valpha(t), t) * Vin(t), right?

Quote:

I went and checked the results.

At $abstime = t1

Vx * V(in)= x1

V(alpha)= y1

At $abstime = t2

Vx * V(in)= x2

V(alpha)= y2

I think y2 should be equal to

y2= [ x2 * (t2 - t1) ] + y1.

I don't think so.

Quote:

But SPECTRE is showing something different than "y2" .

Can anybody please tell where I am going wrong ?
Whether my line of thought is correct ?
or it may be some accuracy issue of Spectre lile Reltol, Vabstol,Abstol playing some rule ?

Also I solve the above differntial eqaution using MATLAB ODE45().
It is totally different from the SPECTRE result !

Did you try an "indirect branch assignment statement"?

- B O E