The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> Gmin
https://designers-guide.org/forum/YaBB.pl?num=1449085758

Message started by Lieutenant Columbo on Dec 2^nd, 2015, 11:49am

Title: Gmin
Post by Lieutenant Columbo on Dec 2^nd, 2015, 11:49am

Is there any reason why someone would want to use a value for gmin other than the default?

Title: Re: Gmin
Post by Ken Kundert on Dec 2^nd, 2015, 12:47pm

Only very rarely, generally on charge storage circuits where voltage droop on capacitors can become a problem.

You need gmin to eliminate the floating-node problem, but once you have it, its value is of little consequence as long as it does not interfere with the circuit. A lot of people change gmin for convergence reasons, but that is a bad practice. You really do not want gmin interfering with the circuit, and it if affects convergence, it is interfering with the circuit.

Thus, you should set gmin to be small enough so that it does not affect the circuit through leakage currents, and address convergence issues in other ways.

-Ken

Title: Re: Gmin
Post by Lieutenant Columbo on Dec 2^nd, 2015, 12:55pm

Thanks.

Is there any issue with having a huge span of R values in a simulation (e.g. 1mOhms to 1TOhms)? Is this a myth?

Title: Re: Gmin
Post by Ken Kundert on Dec 2^nd, 2015, 12:57pm

No. Rather the issue is with large value resistors and conductors. Meaning that if you are using Verilog-A, you should use the resistance form with small valued resistors and the conductance form with large valued resistors. If you use the built-in resistor, this issue it taken care of automatically. The combination of large and small valued resistors is of no consequence.

-Ken

Title: Re: Gmin
Post by Geoffrey_Coram on Dec 2^nd, 2015, 1:35pm

Default GMIN is also a bad idea in JFET circuits, where pA input currents are typical, and thus the same order of magnitude as gmin currents.