The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design Languages >> Verilog-AMS >> Regarding conservative and non-conservative systems
https://designers-guide.org/forum/YaBB.pl?num=1472391699

Message started by AMS_ei on Aug 28^th, 2016, 6:41am

Title: Regarding conservative and non-conservative systems
Post by AMS_ei on Aug 28^th, 2016, 6:41am

Hi,

I am a new with verilog-AMS. I've developed a behavioral model of a lowpass filter.

//-----code begins here-----------//

`include “disciplines.vams”
`include “constants.vams”
module lp_filt_beh (sig1, sig2, gnd);
inout sig1, sig2, gnd;
electrical sig1, sig2, gnd;
parameter real res = 1K;
parameter real cap = 1u;
analog begin
I(sig1,sig2) <+ V(sig1, sig2) / res;
I(sig2,gnd) <+ ddt(V(sig2,gnd) * cap);
end
endmodule

//-------code ends here-----------------//

I am not sure whether it is a conservative system or a non-conservative systems.
How can I distinguish between them from the above example?

Explanation would be great helpful.

Thank you.

Warm regards,

Title: Re: Regarding conservative and non-conservative systems
Post by Ken Kundert on Aug 29^th, 2016, 7:26pm

Systems that satisfy both KCL and KVL are referred to as conservative systems. This is because of Telegan's Theorem, which states that if both KCL and KVL are satisfied, then the system conserves energy. Circuit simulator employ KCL and KVL to formulate their systems, so by definition anything you simulate with a circuit simulator is a conservative system.

However, it is common to refer to models that are written in terms of both voltage and current, as yours is, as conservative models. This distinguishes them from models that are just written in terms of voltage. Such models tend be be more abstract and referred to as signal-flow models.

These are really distinctions without a difference though. Nothing you need to worry about.

-Ken

Title: Re: Regarding conservative and non-conservative systems
Post by Geoffrey_Coram on Aug 30^th, 2016, 11:04am

That's Tellegen's Theorem:
https://en.wikipedia.org/wiki/Tellegen's_theorem

Title: Re: Regarding conservative and non-conservative systems
Post by JayOcad on Sep 8^th, 2016, 12:17am

Thanks for all the guidance on this website and documents- with regards to conservative and non-conservative systems/ signal-flow systems, I had a question about potential signal-flow versus flow signal-flow.

Are there any differences between a signal flow system with a discipline defined with potential Efield only, versus a discipline defined with flow Efield only? If their access, idt, ddt, and all other definitions are exactly the same, does using a "flow" nature differ from using a "potential" nature?

I read in the AMS manuals that in a potential signal-flow system, there is no conservation of charges and, and no sum of *flows* condition (KFL not enforced), but is there a sum of potentials condition in a potential signal-flow system? i.e. If I model Electric field with the "potential" nature declaration, and I need it to disregard conservation/ KPL, is that valid? In a circuit loop with device d1, d2, d3.., does the solver/Spectre try to enforce that the sum of the Electric field across d1,d2,d3.. = 0?

Conversely in a flow signal-flow system, where no potential is defined, I read there is no condition on sum of potentials, does this mean there is neither KPL nor KFL when Spectre sees that only a flow nature is defined? Or is only KPL not enforced and KFL is still enforced(and then Efield is a flow nature, then the Efield into a node must equal the Efield out of a node?).

Thanks for the help,
Best regards,
J

Title: Re: Regarding conservative and non-conservative systems
Post by Ken Kundert on Sep 9^th, 2016, 12:00am

The only difference between full conservative descriptions and signal flow descriptions is that the models written with signal flow descriptions do not have access to the missing nature. Thus, if you use voltage signal flow, the model can neither observe or specify current. If you use current signal flow, the model can neither observe of specify voltage. Other than this there is no difference. That is why you can freely interconnect conservative and signal flow models.

With signal flow models there is some opportunity to simplify the formulation and therefore speed the simulator up a bit, but it is my understanding that there are no simulators that do that.

-Ken

Title: Re: Regarding conservative and non-conservative systems
Post by AMS_ei on Jan 18^th, 2017, 7:57am

Hi Ken,

Thank you for the explanation.
How will you decide whether your design should be of modeled in conservative or in signal flow?

Thank you.

Kind regards,

Title: Re: Regarding conservative and non-conservative systems
Post by Ken Kundert on Jan 18^th, 2017, 4:38pm

Quote:

These are really distinctions without a difference though. Nothing you need to worry about.