The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Modeling >> Behavioral Models >> Verilog-A: the meaning of gnd node?
https://designers-guide.org/forum/YaBB.pl?num=1161265487

Message started by Pavel on Oct 19^th, 2006, 6:44am

Title: Verilog-A: the meaning of gnd node?
Post by Pavel on Oct 19^th, 2006, 6:44am

Hello

Simulating my schema in Virtuoso-AMS I observed non-desirable spikes on some signals.
I've tried to supress them with capacitors:

capacitor #(.c(30p)) C1 (OutP);
capacitor #(.c(30p)) C2 (OutN);
capacitor #(.c(40p)) C3 (DacBuf_InP);
capacitor #(.c(40p)) C4 (DacBuf_InN);

But if I restart the simulation with modified code, I receive internal error (see attachment).

Then I tried following:

electrical gnd;
ground gnd;

capacitor #(.c(30p)) C1 (OutP, gnd);
capacitor #(.c(30p)) C2 (OutN, gnd);
capacitor #(.c(40p)) C3 (DacBuf_InP, gnd);
capacitor #(.c(40p)) C4 (DacBuf_InN, gnd);

This time simulation runs, but spike don't disappear.

So my question is:

Are the instantiations:
capacitor #(.c(30p)) C1 (OutP);

and

capacitor #(.c(30p)) C1 (OutP, gnd);

identical?

Regards.

Pavel

Title: Re: Verilog-A: the meaning of gnd node?
Post by Geoffrey_Coram on Oct 19^th, 2006, 10:49am

Pavel wrote on Oct 19^th, 2006, 6:44am:

Are the instantiations:
capacitor #(.c(30p)) C1 (OutP);

and

capacitor #(.c(30p)) C1 (OutP, gnd);

identical?

I don't think so. I think the first one doesn't specify a connection for the other side of the capacitor, which would mean it is floating. (Since this would have no dc path to ground, the simulator may put GMIN from the second terminal to ground.)

Title: Re: Verilog-A: the meaning of gnd node?
Post by Pavel on Oct 20^th, 2006, 12:55am

Yes, is true. I checked it with simple example.

module GND_NODE_TEST;
electrical A, B, gnd;
ground gnd;
resistor #(.r(10)) R1 (A, gnd);

analog begin
V(A) <+ 3;
V(B) <+ I(A);
end

endmodule

When I measure V(B) it's equal to -0.3V. When I instantiate R1 without gnd:

resistor #(.r(10)) R1 (A);

compilation passed, but simulation doesn't run and gives initialization error.

Regards.

Title: Re: Verilog-A: the meaning of gnd node?
Post by Geoffrey_Coram on Oct 20^th, 2006, 5:35am

In digital Verilog, leaving a port (terminal) unconnected is well-defined; it's treated as an unknown when an input and as high-impedance when an output.

Since Verilog-A is derived from Verilog, it has to follow the same rules. It's a little strange when it iterfaces with Spice, though, because in Spice, terminals must be connected (generally - sometimes, as in BSIMSOI, there are extra terminals for self-heating that are connected conditionally on parameters).

Title: Re: Verilog-A: the meaning of gnd node?
Post by jbdavid on Oct 22^nd, 2006, 10:43am

In Verilog-A
V(A) is the same as V(A,0)
where 0 is the analog simulators ground node.
If it is NOT behaving this way it is not complying to the LRM.
However your simulator may be using a different name for the ground node than you might expect.

if your simulators ground node is gnd! and you were using gnd in your models, then those are different nodes... and you'll get "floating" nodes
so the problem may have been that ONLY those capacitors were connected to the simulators ground node, so it would be apparently floating.
-
Make sure you are using the SAME ground node as the analog simulator.

Jonathan

Title: Re: Verilog-A: the meaning of gnd node?
Post by Geoffrey_Coram on Oct 23^rd, 2006, 5:01am

jbdavid wrote on Oct 22^nd, 2006, 10:43am:

In Verilog-A
V(A) is the same as V(A,0)
where 0 is the analog simulators ground node.
If it is NOT behaving this way it is not complying to the LRM.

Jonathan -
That's *not* the issue here. I am sure the spice capacitor instantiated per Annex E uses the branch access V(A,B). The question is: if you instantiate something in a Verilog netlist without specifying the connection to a port, what happens to that port? (If you don't specify a parameter override, then the default is used, and the language requires a default in the parameter declaration.)

-Geoffrey