Add MonteCarlo simulation

[ra3xdh]

Monte Carlo simulation is only possible using a Nutmeg script. I propose to add a special simulation type that will automatize Monte Carlo script generation. An example of Monte Carlo implementation with Nutmeg script is attached.

LC_band_pass_filter.sch.gz

[lelleuven]

Dear ra3xdh,

Thanks for sharing this code. To my understanding, this is based on the example code of NGSPICE's Monte Carlo simulation. I tried similar code in a toy model ,where I wrote a simple diode model in verilog-a and transfer it into osdi for the usage in NGSPICE. I modified the code above in your example, and it worked fine with NGSPICE from commandline. However, when I tried to incorporate the same model into qucs-s and run simulation, there is always an error, saying no such model. I attached my code below.

• diodeva.va is the verilog-a model and then diodeva.osdi is created by openvaf.
• When I run diode_osdi.sp in NGSPICE, it can simulate a diode behavior using given parameters.
• If I add the code in diode_osdi_dc_MC.sp into diode_osdi.sp, NGSPICE can do Monte Carlo simulation as expected for one parameter I set.
• if I introduce this model to qucs-s by using diodecustom.lib and use the method provided by your post above (use code in diode_osdi_dc_MC.sp as Nutmeg script), there is always an error, saying no such model diode_sch in this case.

I just wonder if the Monte Carlo simulation doesn't apply to the verilog-a model in qucs-s. Or maybe I made any silly mistake somewhere.

Thanks for your time and help.

Diode.zip

[lelleuven]

Monte Carlo simulation is only possible using a Nutmeg script. I propose to add a special simulation type that will automatize Monte Carlo script generation. An example of Monte Carlo implementation with Nutmeg script is attached.

LC_band_pass_filter.sch.gz

Dear ra3xdh,

Could you kindly check my reply above about this new feature?

Thanks a lot !

[ra3xdh]

I have never tried a combination of Nutmeg script simulation and OSDI models. So, I cannot suggest anything.

[lelleuven]

I have never tried a combination of Nutmeg script simulation and OSDI models. So, I cannot suggest anything.

I see. Thanks. But is there any plan to debug this combination in the future?