Re: Too much time on their hands!

Am 19.03.25 um 23:00 schrieb Martin Brown:

On 19/03/2025 14:32, john larkin wrote:

In one recent case the sim kept stalling. I added a 1K resistor off to
the side, one end grounded and the other end connected to nothing.
That fixed things.

 That is odd. I can imagine adding a 1M resistor between some pair of nodes might take the edge off it.
 

Inductors, especially coupled inductors, are quirky. A little ESR
often helps, but that may be random, like the 1K resistor.

 Resonant tank circuits with high Q can sometimes cause trouble.

Trying to solve the matrix equations ( i.e. calculating the nodal
voltages from the G-Matrix and the current vector ) may result in
divide-by-0 errors when a node has absolutely no conductivity to
other nodes. Therefore, some artificial conductivity is introduced.
Otherwise you could not solve a circuit with 2 capacitors in series
and nothing else. The art is to choose the conductivity so that the
equation system can be solved without div by 0, without changing the
original circuit too much. Different spices may handle that
differently; usually there is an option to set the minimum conductivity
to GND if the automatic choice does not work.
There is no such thing in spice like a node that floats in nirvana.
AFAIR there is also a limit variable for g-matrix elements that are
considered for pivoting during solution of the equation system.
Last time I looked deeper into this was when I tried to compile
Spice 2G6 or so on my new 10 MHz 286 with 70 MB disk under
Interactive Unix. Oh, those 64k segments! Fortran memory management
was done by indexing out of the bounds of a small array.  =:-()
That a student had a 70 MB disk all for himself was quite close
to hubris. The institute VAX had 2 * 300 MB Fujitsu Eagle with one
disk for us mortals.
Gerhard