@Mitch Bailey I am designing a constant gm biasing circuit and I am getting large variation due to different temperatures. Below is the schematic of the testbench

@PramitKumar Pal looking at your code you define pretty high range of the temperature variation. I do not know if it is intentional but maybe it is to much and causes a high variation of gm on the output. Look at the definition.

I must have done a mistake in defining the function. What should be a good range for temperature?

I would rather perform STEP TEMP simulation and would analyze gm for different bias conditions instead of performing Monte Carlo on temperature. If you ask me about the range it depends on application, maybe -40 to 125, or 0 to 80, it is really up to your spec

I altered the tb and swept the temp from -40 - 125C. Does the gm plot look better?

Maybe try to elaborate some criteria to judge if its look better or not. Maybe a good starting point would be kind of metric, which evaluates the maximum deviation respect the nominal value of the gm (evaluated at 27 for example). You could look for the min/max value and then report

|max-norm|/norm *100