Diarmuid Collins

Great inputs. Thanks guys.

What exactly is the correct way to apply NGSPICE_MEAS_PRECISION in the .spiceinit file. I use, among many other variations, the following: setenv NGSPICE_MEAS_PRECISION 12 but I see no change in output resolution.

Ah I see. Works as expected now. Thanks for the help.

Attached .spice file

That's very interesting. When I run options inside the .control section I get what appears to be default values, as shown in the attached. This is strange because I had set the below values: .options VNTOL = 1e-9 .options TRTOL = 5 Attached is the .spice file. From this, it appears ngspice is not accepting my updated values.

To see the options selected, do you mean: .options method = gear .options ABSTOL = 1e-12 .options RELTOL = 1e-4 .options VNTOL = 1e-7 .options TRTOL = 14 .options GMIN = 1e-12 .options interp $ interpolates output data onto fixed time steps on a TSTEP grid .options => From the last command, I dont see options being printed out. As you can see from my code, tran analysis is specified by: tran 1n $&xtsim_var I set time step at 1ns but it is my understanding this is the max time step. In other words,...

Hi all, I need to increase simulator accuracy to enable me to analyse a 120dB |CMRR| spec. I do this by placing the below lines: .options ABSTOL = 1e-12 .options RELTOL = 1e-4 .options VNTOL = 1e-7 .options TRTOL = 14 .options GMIN = 1e-12 I know from commercial tools that setting vntol =1e-7 and trtol=14 in particular will enable you to simulate to 120dB+ levels. However, I also know from commercial tools that this will bring the simulation time to an absolute crawl. Issue is I dont see any of that...

Apologies. I updated this on Element but had not yet got around to updating here. Root cause of the convergence issue was creating an xschem symbol by just copy pasting, as opposed to creating from the root schematic, as is the correct procedure. .