Hi all

I'm sweeping in temperature (DC analysis) a voltage/current reference + comparator based temperature flag (spectre simulator). Unfortunately, most of the corner points are not converging even thought I feed the DC nodeset file from a previous transient simulation.

When the comparator trigger, the DC simulation fails to find the desirable OP point and all the references and currents goes to zero.  

I have tried different homotopy, rforce and gmin options without success.  This is not the first time that it happen to me with this kind of circuits.

Does anyone has any suggestion or comment to improve the simulator convergence in this kind of circuits?

Thanks